added text inputs for 5 protocols

10x_chromium_3_gene_expression_v4/10xChromium3v4.human_text.txt

@@ -0,0 +1,155 @@
+Chromium GEM-X Single Cell 3' Gene Expression v4
+
+Protocol
+
+Chromium GEM-X Single Cell 3' Gene Expression v4 generates dual-index Illumina 3' gene-expression libraries from single cells. Cells are partitioned into GEMs with barcoded GEM-X gel beads. Inside each GEM, polyadenylated mRNA is reverse transcribed using gel bead primers that contain the TruSeq Read 1 sequence, a 16-bp 10x cell barcode, a 12-bp UMI, and a 30-nt poly(dT)VN capture sequence. After GEM-RT, barcoded full-length cDNA is purified, amplified, fragmented, end-repaired, A-tailed, adapter-ligated, and amplified by sample index PCR to generate the final dual-index gene-expression library.
+
+The gel beads also contain Capture Sequence 1 for Feature Barcode-compatible targets, but this Gene Expression-only protocol uses the poly(dT) primers for library generation.
+
+Key oligo and library-related sequences
+
+1. Gel Bead Primer
+Source sequence:
+5'-CTACACGACGCTCTTCCGATCT-N16-N12-TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVN-3'
+
+
+2. Gene Expression Library — Sample Index PCR Product, top strand
+Source sequence:
+5'-AATGATACGGCGACCACCGAGATCTACAC-N10-ACACTCTTTCCCTACACGACGCTCTTCCGATCT-N16-N12-T30-VN-cDNA_Insert-AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC-N10-ATCTCGTATGCCGTCTTCTGCTTG-3'
+
+
+3. Gene Expression Library — Sample Index PCR Product, bottom strand
+Source sequence:
+3'-TTACTATGCCGCTGGTGGCTCTAGATGTG-N10-TGTGAGAAAGGGATGTGCTGCGAGAAGGCTAGA-N16-N12-A30-BN-cDNA_Insert-TCTAGCCTTCTCGTGTGCAGACTTGAGGTCAGTG-N10-TAGAGCATACGGCAGAAGACGAAC-5'
+
+
+
+Step-by-step library generation
+
+Step 1. GEM generation and reverse transcription
+
+Input substrate:
+polyadenylated mRNA from a single cell
+
+Added oligos/reagents:
+- GEM-X gel bead primer
+- Template Switch Oligo B
+- reverse-transcription reagents
+
+Molecular event:
+The gel bead primer captures polyadenylated mRNA through its poly(dT)VN region. Reverse transcription creates barcoded full-length cDNA. Each cDNA molecule receives a common cell barcode from the GEM and a molecule-specific UMI.
+
+Product structure:
+
+TruSeq Read 1 partial sequence
++ CELL_BARCODE
++ UMI
++ T30VN
++ cDNA insert
++ TSO-derived sequence
+
+
+Step 2. Post-GEM cleanup and cDNA amplification
+
+Input substrate:
+- barcoded first-strand cDNA from GEM-RT
+
+Added oligos/reagents:
+- cDNA Primers
+- Amp Mix
+
+Molecular event:
+
+GEMs are broken, first-strand cDNA is purified, and barcoded full-length cDNA is amplified to generate sufficient material for library construction.
+
+Product structure:
+- amplified barcoded full-length cDNA
+
+Conceptual structure:
+
+TruSeq Read 1 partial sequence
++ CELL_BARCODE
++ UMI
++ T30VN
++ cDNA insert
++ TSO-derived sequence
+
+Step 3. Fragmentation, end repair, and A-tailing
+
+Input substrate:
+- amplified barcoded full-length cDNA
+
+Added reagents:
+- Fragmentation Buffer
+- Fragmentation Enzyme
+
+Molecular event:
+
+cDNA is enzymatically fragmented, end-repaired, and A-tailed to prepare fragments for adapter ligation.
+
+Product structure:
+- barcoded cDNA fragment with repaired and A-tailed ends
+
+Conceptual structure:
+
+TruSeq Read 1 partial sequence
++ CELL_BARCODE
++ UMI
++ T30VN
++ cDNA fragment
+
+Step 4. Adapter ligation
+
+Input substrate:
+- fragmented, end-repaired, A-tailed cDNA
+
+Added oligos/reagents:
+- Adaptor Ligation Mix
+- DNA Ligase
+
+Molecular event:
+
+Sequencing adapter sequence is ligated to the cDNA fragment, introducing the region that will become the Read 2 side of the final library.
+
+Product structure:
+
+TruSeq Read 1 partial sequence
++ CELL_BARCODE
++ UMI
++ T30VN
++ cDNA fragment
++ Read 2-side adapter sequence
+
+Step 5. Sample index PCR
+
+Input substrate:
+- adapter-ligated cDNA fragments
+
+Added oligos/reagents:
+- Dual Index TT Set A
+- Library Amp Mix or Amp Mix
+
+Molecular event:
+Sample index PCR adds Illumina P5, Illumina P7, the i5 sample index, the i7 sample index, and completes the final sequencing-ready dual-index library.
+
+
+
+Final canonical library structure
+
+Simplified segment-level structure:
+P5 + i5 index + TruSeq Read 1 + 10x cell barcode + UMI + poly(dT)VN + cDNA insert + TruSeq Read 2 + i7 index + P7
+
+Sequencing read interpretation:
+- Read 1: 16-bp 10x cell barcode + 12-bp UMI
+- Read 2: cDNA insert
+- i5 index read: 10-bp i5 sample index
+- i7 index read: 10-bp i7 sample index
+
+Human-curation notes
+
+1. The appendix explicitly provides the Gel Bead Primer and Gene Expression Library Sample Index PCR Product sequences.
+2. The protocol states that GEM-X gel beads also include Capture Sequence 1 for Feature Barcode-compatible targets, but only the poly(dT) primers are used in this Gene Expression-only protocol. Therefore Capture Sequence 1 should not be included in the GEX final library structure.
+3. Template Switch Oligo B is listed as a reagent and appears conceptually in the diagrams, but its full sequence is not explicitly provided in the oligonucleotide sequence appendix. It should not be assigned an exact sequence unless completed from an approved memory source.
+4. cDNA Primers are listed as kit reagents, but their exact sequences are not explicitly provided in this document. They should not be included as exact oligo-sequence ground truth unless completed from approved memory.
+5. The source uses N10 on both sides of the final library. In this dual-index protocol, these correspond to 10-bp i5 and i7 sample indexes.
+6. The source bottom strand is written 3'→5'. The canonical final library structure for scoring should use the top strand written 5'→3'.

10x_chromium_single_cell_atac_v2/10xChromium_scATACv2.human_text.txt

@@ -0,0 +1,163 @@
+Chromium Next GEM Single Cell ATAC Reagent Kits v2
+
+Protocol
+Chromium Next GEM Single Cell ATAC Reagent Kits v2 profiles chromatin accessibility from single nuclei. Nuclei are first transposed in bulk. During transposition, transposase enters nuclei, fragments accessible genomic DNA, and adds adapter sequences to the ends of DNA fragments. Transposed nuclei are then partitioned into GEMs with barcoded gel beads. Inside each GEM, gel bead oligonucleotides containing an Illumina P5 sequence, a 16-nt 10x barcode, and a Read 1N sequence are released and used to generate 10x-barcoded DNA fragments. After GEM incubation, barcoded DNA fragments are purified, unused barcodes are removed, and sample index PCR adds P7 and an i7 sample index to generate the final Illumina-ready Single Cell ATAC library.
+
+The final library is a paired-end, single-cell ATAC library. Read 1N and Read 2N sequence the genomic insert from opposite ends of the transposed fragment. The i7 index read contains the 8-bp sample index, and the i5 index read contains the 16-bp 10x barcode.
+
+Key oligo and library-related sequences
+
+The explicit sequence strings are provided in the A1 Oligonucleotide Sequences appendix.
+
+1. Transposed DNA Product
+Source sequence:
+5'-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG-----insert--CTGTCTCTTATACACATCT-3'
+3'-TCTACACATATTCTCTGTC--insert-----GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-5'
+
+2. Gel Bead Oligo Primer, PN-2000210
+Source sequence:
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTC-3'
+
+3. Linear Amplification DNA Product
+Source sequence:
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG----insert----CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-3'
+
+4. SI-PCR Primer B, PN-2000128, and Single Index Plate N Set A, PN-3000427
+Forward Primer:
+5'-AATGATACGGCGACCACCGAGA-3'
+
+Reverse Primer:
+5'-CAAGCAGAAGACGGCATACGAGAT-NNNNNNNN-GTCTCGTGGGCTCGG-3'
+
+5. Sample Index PCR Product
+Source sequence:
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG---insert---CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-NNNNNNNN-ATCTCGTATGCCGTCTTCTGCTTG-3'
+3'-TTACTATGCCGCTGGTGGCTCTAGATGTG-NNNNNNNNNNNNNNNN-AGCAGCCGTCGCAGTCTACACATATTCTCTGTC---insert---GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-NNNNNNNN-TAGAGCATACGGCAGAAGACGAAC-5'
+
+
+
+Step-by-step library generation
+
+Step 1. Transposition
+
+Input substrate:
+isolated nuclei / genomic DNA in accessible chromatin
+
+Added oligos/reagents:
+
+* ATAC Buffer B
+* ATAC Enzyme B / transposase
+
+Molecular event:
+Transposase enters nuclei and fragments DNA in open chromatin. At the same time, adapter sequences are added to the ends of the DNA fragments.
+
+Product structure:
+
+Read 1N-side transposase adapter
+
+* genomic DNA insert
+* Read 2N-side transposase adapter
+
+Step 2. GEM generation and barcoding
+
+Input substrate:
+
+* transposed nuclei
+* transposed DNA fragments
+
+Added oligos/reagents:
+
+* Single Cell ATAC Gel Beads v2
+* Barcoding Reagent B
+* Barcoding Enzyme
+* Reducing Agent B
+* Partitioning Oil
+
+Molecular event:
+Transposed nuclei are partitioned into GEMs. Gel bead oligos containing P5, a 16-nt 10x barcode, and a partial Read 1N sequence are released. During GEM incubation, thermal cycling produces 10x-barcoded single-stranded DNA fragments.
+
+Product structure:
+
+P5 partial sequence
+
+* 10x barcode
+* Read 1N sequence
+* genomic DNA insert
+* Read 2N-side sequence
+
+Step 3. Post-GEM incubation cleanup
+
+Input substrate:
+
+* 10x-barcoded DNA fragments from GEM incubation
+
+Added oligos/reagents:
+
+* Recovery Agent
+* Dynabeads MyOne SILANE
+* Cleanup Buffer
+* SPRIselect reagent
+
+Molecular event:
+GEMs are broken, biochemical reagents are removed, and unused barcodes are depleted.
+
+Product structure:
+
+* purified 10x-barcoded ATAC DNA fragments
+
+Conceptual structure:
+
+P5 partial sequence
+
+* 10x barcode
+* Read 1N sequence
+* genomic DNA insert
+* Read 2N-side sequence
+
+Step 4. Sample index PCR
+
+Input substrate:
+
+* purified 10x-barcoded ATAC DNA fragments
+
+Added oligos/reagents:
+
+* SI-PCR Primer B
+* Single Index Plate N Set A
+* Amp Mix
+
+Molecular event:
+Sample index PCR adds the P7 side and the 8-bp i7 sample index, completing the final Illumina-ready Single Cell ATAC library.
+
+Product structure:
+
+P5
+
+* 10x barcode
+* Read 1N
+* genomic DNA insert
+* Read 2N
+* sample index
+* P7
+
+Final canonical library structure
+
+Simplified segment-level structure:
+
+P5 + 10x cell barcode + Read 1N + genomic DNA insert + Read 2N + i7 sample index + P7
+
+Sequencing read interpretation:
+
+* Read 1N: genomic insert sequence from one end of the ATAC fragment
+* Read 2N: genomic insert sequence from the opposite end of the ATAC fragment
+* i7 index read: 8-bp sample index
+* i5 index read: 16-bp 10x barcode
+
+Human-curation notes
+
+1. The appendix explicitly provides the Transposed DNA Product, Gel Bead Oligo Primer, Linear Amplification DNA Product, SI-PCR Primer B forward and reverse primers, and Sample Index PCR Product sequences.
+2. The protocol states that gel bead oligonucleotides contain P5, a 16-nt 10x barcode, and Read 1N sequence.
+3. The protocol states that P7 and a sample index are added during library construction by sample index PCR.
+4. The document does not print individual per-well sample index sequences. The sample index is shown as an 8-bp N-run in the reverse primer and final Sample Index PCR Product.
+5. The source provides both top and bottom strands. The bottom strand is written 3'→5'. The canonical library structure should be interpreted using the top strand written 5'→3'.
+6. The 16-bp 10x barcode is part of the final library and is read through the i5 index read, while the 8-bp sample index is read through the i7 index read.

10x_chromium_single_cell_atac_v2/10xChromium_scATACv2.mineru_ocr.txt

@@ -1,6 +1,6 @@
 # MinerU OCR text conversion
 source_pdf: 10xChromium_scATACv2.pdf
-mineru_command: mineru -m ocr -b pipeline -f false -t false
+mineru_command: mineru -m ocr -b pipeline -f false -t false; appendix supplement from mineru -b hybrid-engine --effort high --image-analysis true -m ocr -s 51 -e 51 -f false -t false
 
 ![](images/15db7fcbdaadc3c4ce127ef8f0e092327aff2adad7473c0750c58760b85294d1.jpg)
 
@@ -1030,4 +1030,78 @@ Protocol Step 2.5 – GEM Incubation
 5\*-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG----inSert----CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-3'  
 Protocol Step 4.1 – Sample Index PCR  
 ![](images/b7d86753aa03d60ceef5d984f3cd0deb9707f8f490527bb2d5ef35fb34979611.jpg)  
-5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG--InSer---CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-NNNNNNNN-ATCTCGTATGCCGTCTTCTGCTTG-3 3'-TTACTATGCCGCTGGTGGCTCTAGATGTG-NNNNNNNNNNNNNNNN-AGCAGCCGTCGCAGTCTACACATATTCTCTGTC---InSert--GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-NNNNNNNN-TAGAGCATACGGCAGAAGACGAAC-5"
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG--InSer---CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-NNNNNNNN-ATCTCGTATGCCGTCTTCTGCTTG-3 3'-TTACTATGCCGCTGGTGGCTCTAGATGTG-NNNNNNNNNNNNNNNN-AGCAGCCGTCGCAGTCTACACATATTCTCTGTC---InSert--GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-NNNNNNNN-TAGAGCATACGGCAGAAGACGAAC-5"
+
+## MinerU hybrid high OCR appendix supplement
+source_page: 52
+source_tmp_output: /private/tmp/mineru_atac_hybrid_high_page52/10xChromium_scATACv2/hybrid_ocr/10xChromium_scATACv2.md
+mineru_command: mineru -p 10xChromium_scATACv2.pdf -o /private/tmp/mineru_atac_hybrid_high_page52 -b hybrid-engine --effort high --image-analysis true -m ocr -s 51 -e 51 -f false -t false
+
+## A1 Oligonucleotide Sequences
+
+Protocol steps correspond to the Chromium Next GEM Single Cell ATAC Reagent Kits v2 User Guide (CG000496)  
+Protocol Step 1 – Transposition  
+Transposed DNA Product  
+![](images/0e5168edefbddeae1c35e8b8b597f65941d6f8d9171d0d500bbb7c6607e0df58.jpg)  
+5'-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG----insert-- CTGTCTCTTATACACATCT-3'
+3'-TCTACACATATTCTCTGTC --insert----GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-5'
+
+Protocol Step 2.5 – GEM Incubation  
+Gel Bead Oligo
+Primer
+PN-2000210  
+![](images/f0187dc1af85bd8abdad2e978938e166853f6e48f938ea6e10caa9900d4c611c.jpg)  
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTC-3'
+
+Linear Amplification
+DNA Product  
+![](images/81758b0c81180cfd89f2c7fe36192196b2ca45a56ff4b24ebfd8710157dbb0e8.jpg)
+
+<details>
+<summary>text_image</summary>
+
+P5
+10x
+Barcode
+Read 1N
+Insert
+Read 2N
+</details>
+
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG----insert----CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-3'
+
+Protocol Step 4.1 – Sample Index PCR  
+SI-PCR Primer B
+PN-2000128  
+![](images/6ed8066003d982150987e8f58a0bf8ab52f48a5e6c2f4b435722a14b883e6e88.jpg)  
+5'-AATGATACGGCGACCACCGAGA-3'
+
+![](images/896a8bda8a834dfc5b78ae3b5b71ac6d93918f2339952073dce0cd65fbd0a08c.jpg)
+
+<details>
+<summary>text_image</summary>
+
+Reverse Primer:
+P7
+Sample
+Index N
+Partial
+Read 2N
+5'- CAAGCAGAAGACGGCATACGAGAT-NNNNNNNN-GTCTCGTGGGCTCGG-3'
+</details>
+
+Single Index Plate N
+Set A
+PN-3000427  
+Sample Index PCR
+Product  
+![](images/eb0ebef64524a7b98d9a9c18bc51334dbb468265363980e7b176ebbe9e713ebf.jpg)
+
+<details>
+<summary>text_image</summary>
+
+P5 10x Barcode Read 1N Insert Read 2N Sample Index N P7
+</details>
+
+5'-AATGATACGGCGACCACCGAGATCTACAC-NNNNNNNNNNNNNNNNN-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG---insert---CTGTCTCTTATACACATCTCCGAGCCCACGAGAC-NNNNNNNNN-ATCTCGTATGCCGTCTTCTGCTTG-3'
+3'-TTACTATGCCGCTGGTGGCTCTAGATGTG-NNNNNNNNNNNNNNNNN-AGCAGCCGTCGCAGTCTACACATATTCTCTGTC---insert---GACAGAGAATATGTGTAGAGGCTCGGGTGCTCTG-NNNNNNNNN-TAGAGCATACGGCAGAAGACGAAC-5'

cel_seq/CEL-Seq.human_text.txt

@@ -0,0 +1,430 @@
+CEL-seq
+
+Protocol
+
+CEL-seq, Cell Expression by Linear amplification and Sequencing, is a multiplexed single-cell RNA-seq method. Individual cells are reverse transcribed with uniquely barcoded anchored polyT primers. Each RT primer contains a T7 promoter, an Illumina 5' adapter sequence, an 8-bp cell barcode, and an anchored polyT sequence. After second-strand synthesis, barcoded cDNA samples are pooled and linearly amplified by in vitro transcription. The amplified RNA is fragmented, processed using a modified Illumina directional RNA library preparation workflow, ligated to the Illumina 3' adapter, reverse transcribed, and PCR amplified. The final library is sequenced paired-end: Read 1 recovers the cell barcode and polyT stretch, while Read 2 identifies the mRNA transcript.
+
+Key oligo and library-related sequences
+
+The explicit CEL-seq RT primer sequences are provided in the protocol. Each primer has the same shared architecture:
+
+T7 promoter + Illumina 5' adapter + 8-bp cell barcode + anchored polyT sequence
+
+Shared CEL-seq RT primer prefix contains the T7 promoter followed by the Illumina 5' adapter sequence. It is followed by an 8-bp barcode and then a 24T+V anchored polyT region.
+
+1. CEL-seq RT primer #1
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+2. CEL-seq RT primer #2
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+3. CEL-seq RT primer #3
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+4. CEL-seq RT primer #4
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+5. CEL-seq RT primer #5
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAATCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+6. CEL-seq RT primer #6
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTATCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+7. CEL-seq RT primer #7
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACATCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+8. CEL-seq RT primer #8
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGATCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+9. CEL-seq RT primer #9
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCATCCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+10. CEL-seq RT primer #10
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+11. CEL-seq RT primer #11
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+12. CEL-seq RT primer #12
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+13. CEL-seq RT primer #13
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAGAATTTTTTTTTTTTTTTTTTTTTTTTV
+
+14. CEL-seq RT primer #14
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTGAATTTTTTTTTTTTTTTTTTTTTTTTV
+
+15. CEL-seq RT primer #15
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACGAATTTTTTTTTTTTTTTTTTTTTTTTV
+
+16. CEL-seq RT primer #16
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGGAATTTTTTTTTTTTTTTTTTTTTTTTV
+
+17. CEL-seq RT primer #17
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+18. CEL-seq RT primer #18
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+19. CEL-seq RT primer #19
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+20. CEL-seq RT primer #20
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+21. CEL-seq RT primer #21
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACATCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+22. CEL-seq RT primer #22
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+23. CEL-seq RT primer #23
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+24. CEL-seq RT primer #24
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+25. CEL-seq RT primer #25
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+26. CEL-seq RT primer #26
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+27. CEL-seq RT primer #27
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+28. CEL-seq RT primer #28
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+29. CEL-seq RT primer #29
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+30. CEL-seq RT primer #30
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+31. CEL-seq RT primer #31
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+32. CEL-seq RT primer #32
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+
+33. CEL-seq RT primer #33
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+34. CEL-seq RT primer #34
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+35. CEL-seq RT primer #35
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCATCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+36. CEL-seq RT primer #36
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+
+37. CEL-seq RT primer #37
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+38. CEL-seq RT primer #38
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+39. CEL-seq RT primer #39
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+40. CEL-seq RT primer #40
+
+Source sequence:
+
+CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+Step-by-step library generation
+
+Step 1. Single-cell isolation and reverse transcription
+
+Input substrate:
+polyadenylated mRNA from an individual cell
+
+Added oligos/reagents:
+
+* one uniquely barcoded CEL-seq RT primer
+* reverse-transcription reagents
+* optional ERCC spike-in mix
+
+Molecular event:
+The CEL-seq RT primer anneals to the polyA tail through its anchored polyT region. Reverse transcription creates first-strand cDNA. Each cell receives a unique 8-bp barcode through the RT primer. The RT primer also contributes the T7 promoter and Illumina 5' adapter sequence.
+
+Product structure:
+
+T7 promoter
+
+* Illumina 5' adapter
+* cell barcode
+* anchored polyT
+* cDNA from mRNA
+
+Step 2. Second-strand synthesis and pooling
+
+Input substrate:
+
+* first-strand cDNA from individual cells
+
+Added oligos/reagents:
+
+* second-strand synthesis reagents from the MessageAmp II kit
+
+Molecular event:
+Second-strand synthesis creates double-stranded cDNA. After this step, barcoded cDNA reactions from multiple cells are pooled before in vitro transcription.
+
+Product structure:
+
+* pooled, barcoded double-stranded cDNA containing T7 promoter sequence
+
+Step 3. In vitro transcription linear amplification
+
+Input substrate:
+
+* pooled, barcoded double-stranded cDNA
+
+Added oligos/reagents:
+
+* T7 IVT reagents from the MessageAmp II kit
+
+Molecular event:
+The T7 promoter drives in vitro transcription, generating linearly amplified RNA from the pooled barcoded cDNA templates.
+
+Product structure:
+
+* amplified RNA carrying the cell barcode and 5' adapter-derived sequence
+
+Step 4. RNA fragmentation
+
+Input substrate:
+
+* amplified RNA
+
+Added reagents:
+
+* fragmentation buffer
+* fragmentation stop buffer
+
+Molecular event:
+The amplified RNA is fragmented to a size range appropriate for sequencing.
+
+Product structure:
+
+* fragmented amplified RNA
+
+Step 5. End repair / phosphatase and PNK treatment
+
+Input substrate:
+
+* fragmented amplified RNA
+
+Added reagents:
+
+* Antarctic phosphatase
+* PNK
+* ATP
+* RNase inhibitor
+
+Molecular event:
+Fragmented RNA is enzymatically processed to make ends compatible with adapter ligation.
+
+Product structure:
+
+* end-repaired fragmented RNA
+
+Step 6. 3' adapter ligation
+
+Input substrate:
+
+* end-repaired fragmented RNA
+
+Added oligos/reagents:
+
+* diluted Illumina RA3 3' adapter
+* T4 RNA Ligase 2, truncated
+* ligation buffer
+* stop solution
+
+Molecular event:
+The Illumina 3' adapter is ligated to fragmented RNA.
+
+Product structure:
+
+* adapter-ligated fragmented RNA
+
+Step 7. Reverse transcription
+
+Input substrate:
+
+* 3' adapter-ligated fragmented RNA
+
+Added oligos/reagents:
+
+* Illumina RNA RT Primer
+* reverse-transcription reagents
+
+Molecular event:
+Adapter-ligated RNA is reverse transcribed into cDNA.
+
+Product structure:
+
+* cDNA library molecule containing the CEL-seq barcode side and Illumina 3' adapter side
+
+Step 8. PCR amplification
+
+Input substrate:
+
+* reverse-transcribed cDNA library molecule
+
+Added oligos/reagents:
+
+* RNA PCR Primer RP1
+* indexed RNA PCR Primer RPIX
+* PCR mix
+
+Molecular event:
+PCR selects and amplifies molecules containing both Illumina adapters. Indexed RNA PCR primer can add a library-level Illumina index from the TruSeq small RNA kit.
+
+Product structure:
+
+* final CEL-seq Illumina sequencing library
+
+Final canonical library structure
+
+Simplified segment-level structure:
+
+T7 promoter / Illumina 5' adapter + cell barcode + polyT stretch + mRNA-derived insert + Illumina 3' adapter + Illumina PCR/index sequences
+
+Sequencing read interpretation:
+
+* Read 1: cell barcode followed by polyT stretch
+* Read 2: mRNA transcript-derived sequence
+* Optional Illumina index: library-level index from the indexed RNA PCR primer, if used
+
+Human-curation notes
+
+1. The paper describes CEL-seq as using a reverse-transcription primer containing an anchored polyT, unique barcode, Illumina 5' adapter, and T7 promoter.
+2. The protocol explicitly lists 40 CEL-seq RT primer sequences.
+3. The shared RT primer architecture is T7 promoter + Illumina 5' adapter + 8-bp barcode + 24T+V anchored polyT.
+4. After second-strand synthesis, barcoded cDNA reactions are pooled before in vitro transcription.
+5. The amplified RNA is fragmented, processed, ligated to the Illumina 3' adapter, reverse transcribed, and PCR amplified.
+6. The protocol names Illumina small-RNA kit oligos such as RA3, RNA RT Primer, RNA PCR Primer RP1, and indexed RNA PCR Primer RPIX, but it does not print their sequences. Their exact nucleotide sequences should not be added from this input alone.
+7. The source-visible CEL-seq library segment is the CEL-seq primer-derived region containing the Illumina 5' adapter, 8-bp cell barcode, and anchored polyT region. Downstream Illumina small-RNA kit adapter, index, and P7 sequences are not printed in the provided protocol text.
+8. The paper states that paired-end sequencing is used: the first read recovers the barcode, whereas the second read identifies the mRNA transcript.

cel_seq/CEL-Seq_paper.docling_text.txt

@@ -0,0 +1,642 @@
+# docling text extraction
+source_pdf: CEL-Seq_paper.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/cel_seq/CEL-Seq_paper.pdf
+extraction: Docling: PdfPipelineOptions(do_ocr=False); assembled markdown plus Docling native PDF layout prediction text cells sorted by page/y/x
+
+## Docling assembled markdown
+
+<!-- image -->
+
+## CEL-Seq: Single-Cell RNA-Seq by Multiplexed Linear Amplification
+
+Tamar Hashimshony, 1,2 Florian Wagner, 1,2 Noa Sher, 1,2 and Itai Yanai 1, * 1 Department of Biology, Technion-Israel Institute of Technology, Haifa 32000, Israel 2 These authors contributed equally to this work *Correspondence: yanai@technion.ac.il http://dx.doi.org/10.1016/j.celrep.2012.08.003
+
+## SUMMARY
+
+High-throughput sequencing has allowed for unprecedented detail in gene expression analyses, yet its efficient application to single cells is challenged by the small starting amounts of RNA. We have developed CEL-Seq, a method for overcoming this limitation by barcoding and pooling samples before linearly amplifying mRNA with the use of one round of in vitro transcription. We show that CEL-Seq gives more reproducible, linear, and sensitive results than a PCR-based amplification method. We demonstrate the power of this method by studying early C. elegans embryonic development at single-cell resolution. Differential distribution of transcripts between sister cells is seen as early as the two-cell stage embryo, and zygotic expression in the somatic cell lineages is enriched for transcription factors. The robust transcriptome quantifications enabled by CEL-Seq will be useful for transcriptomic analyses of complex tissues containing populations of diverse cell types.
+
+## INTRODUCTION
+
+For many biological questions a single-cell-level description of gene regulation is advantageous to cell populations (Tang et al., 2011; Wang and Bodovitz, 2010). Microscopy, FACS, or real-time PCR-based methods can provide a single-cell aspect to experiments but are able to assay only a handful of genes at a time. High-throughput technologies such as microarrays and RNA-Seq provide a full view of the expression of all genes but are limited by the amount of RNA needed for analysis. This can be solved by adding an RNA amplification step, either by exponential PCR-based amplification or linear in vitro transcription (IVT) amplification (Eberwine et al., 1992). With PCR practically any RNA starting amount can be employed, simply by adding additional cycles, thereby allowing analysis at the single-cell level. However, efforts for linear amplification of RNA from single cells have been challenged by IVT's lower bound of  400 pg total RNA as input material for a single round of amplification. Therefore, to date, IVT has not been efficiently used for amplification of RNA from single cells (Tang et al., 2011).
+
+With the dramatically decreasing costs of sequencing, RNASeq (Wang et al., 2009) has emerged as the preferred method for transcriptomic analyses, overtaking microarrays, providing an imperative for any transcriptomic method to be adapted for RNA-Seq. A PCR-based amplification protocol has been used in combination with SOLID sequencing (Tang et al., 2009). Recently, the PCR-based method has been extended to include a multiplexing step for the amplification of multiple cells in parallel, allowing for high-throughput analysis, and uses the Illumina sequencing platform (Islam et al., 2011). In comparison, IVT of single cells has been described before; however, it is labor intensive (Eberwine et al., 1992), requiring three rounds of amplification (  5 days work/cell) and has not been adapted for multiplexed sequencing. These considerations have hitherto prevented the higher quality possible with IVT from being adapted for single-cell RNA-Seq.
+
+Here, we present CEL-Seq (Cell Expression by Linear amplification and Sequencing), a protocol that meets the demand of linear amplification by IVT for sufficient material by pooling barcoded samples, therefore allowing the efficient linear amplification of RNA from single cells and their analysis by sequencing. Wecompare the performance of our method on two mammalian cell types to that of a PCR-based approach and use spike-ins to establish CEL-Seq's exact reproducibility and sensitivity at very low amounts of input RNA. Finally, we apply our protocol to study sister cells from early C. elegans embryos, and demonstrate that CEL-Seq's high performance can be used to reliably distinguish between cell types, even in cases where only subtle biological differences are present.
+
+## RESULTS
+
+## CEL-Seq Performs Multiplexed Single-Cell Transcriptomics by Linear Amplification
+
+The CEL-Seq method begins with a single-cell reverse-transcription reaction using a primer designed with an anchored polyT, a unique barcode, the 5 0 Illumina sequencing adaptor, and a T7 promoter (Figure 1A; see Experimental Procedures for details). Next, second-strand synthesis is performed and then the cDNA samples are pooled and consequently comprise sufficient template material for an IVT reaction. The amplified RNA is then subjected to directional RNA library preparation. The RNA is fragmented to a size distribution appropriate for sequencing, the Illumina 3 0 adaptor is added by ligation, RNA is reverse transcribed to DNA, and the 3 0 -most fragments that
+
+<!-- image -->
+
+<!-- image -->
+
+## Figure 1. The CEL-Seq Method
+
+(A) Individual cells are added to tubes, each with a uniquely bar-coded primer for reverse transcription. After second-strand synthesis, the reactions are pooled for IVT. The amplified RNA is then fragmented and purified before entry into a modified version of the Illumina directional RNA protocol, the molecules with both Illumina adaptors are selected, and the DNA library is sequenced with paired-end reads.
+
+- (B) Nucleotide distribution in the sequenced paired-end reads. Each nucleotide position is represented by one column, with the first base on the left.
+- (C) Barcode distribution of one IVT reaction after demultiplexing. The cells from three two-cell stage C. elegans embryos (denoted P1 and AB) and a single one-cell stage embryo (denoted P0) were amplified together in a single multiplexed IVT reaction.
+- (D) Distribution of the reads mapping to the C. elegans genome in the six AB/P1 cells. Error bars indicate the SD.
+
+(E) Correlation between biological AB replicates.
+
+See also Figure S1C.
+
+contain both Illumina adaptors and a barcode are selected. The resulting library undergoes paired-end sequencing, where the first read recovers the barcode, whereas the second identifies the mRNA transcript (Figure 1A). Thus, by multiplexing CELSeq takes advantage of the different input requirements of the reverse-transcription and IVT reactions to obtain sufficient RNA from single cells for a single round of linear amplification.
+
+As an initial test, we applied CEL-Seq to individual cells isolated from three two-cell C. elegans embryos (Table S1). On average, 95.5% of the filtered reads had a barcode located precisely at the beginning of the first read, invariably followed by a polyT stretch (Figure 1B). Barcodes from all six samples were represented, indicating the success of the individual single-cell reverse-transcription reactions (Figure 1C). We mapped reads to the C. elegans genome and found that 91.7% stemmed from mRNA. Only 2.0% stemmed from ribosomal RNA(rRNA), demonstrating CEL-Seq's specificity for polyadenylated transcripts. This was also supported by the extremely lowdetected expression levels of core histone mRNAs, which are thought to be highly expressed yet mostly nonpolyadenylated
+
+<!-- image -->
+
+<!-- image -->
+
+<!-- image -->
+
+(A) CEL-Seq Pearson's correlation coefficients among the ES and MEF cells (on log10 tpm values), computed as previously described by Islam et al. (2011), on the 1,000 genes most highly expressed in ES cells and 1,000 genes most highly expressed in MEF cells (1,385 genes).
+
+(B) Mean number of genes detected above two thresholds (10 and 100 tpm) across the ES and MEF cell types using CEL-Seq and the ''STRT'' PCR-based
+
+(C) Reproducibility according to expression level. For each gene the coefficient of variation was computed across the log10 tpm values in the ES cells for the STRT and CEL-Seq methods. The genes were then ranked by expression level in bins of 200 from high to low. For each bin the mean and SD of the coefficients of
+
+Figure 2. Benchmarking of CEL-Seq on Mouse ES and MEF Cells method (Islam et al., 2011). Error bars indicate 95% confidence intervals. variation of the genes are shown.
+
+See also Figure S2 for additional analyses.
+
+(Figure S1A). Finally, RNase treatment of cells did not produce amplified RNA indicating the specificity of the method to RNA.
+
+CEL-Seq is highly strand specific because 97.2% of exonic reads exhibited sense orientation (Figure 1D). The reads mapped exclusively to the 3 0 end of transcripts (Figure S1B), which was expected because CEL-Seq only retains the 3 0 -most fragments of transcripts (Figure 1A). Some reads mapped to intergenic sequences, but manual inspection of the aligned reads revealed that this can likely be explained by incomplete 3 0 UTR annotations (data not shown). Expression levels were then estimated by counting all reads mapping to each gene, and normalized to give the read count in transcripts per million (tpm; see Experimental Procedures). The expression levels of all genes (henceforth, transcriptome) across biological replicates showed an average correlation of R = 0.979 (Figures 1E and S1C). A negative control-starting with the growth media lacking a cellresulted in very few reads (Figure S1D).
+
+## CEL-Seq Outperforms a PCR-Based Multiplexed RNASeq Method
+
+We next sought to compare the performance of our protocol to the STRT method, a previously introduced PCR-based multiplexed RNA-Seq method by Islam et al. (2011). We thus applied CEL-Seq to the cell types compared in this previous study by Islam et al. (2011) and determined the transcriptomes of nine mouse embryonic stem (ES) cells and seven mouse embryonic fibroblasts (MEFs) (Table S1). When comparing the distribution of expression levels of each single-cell transcriptome across methods and cell type, CEL-Seq shows more reproducible distributions of expression (Figure S2A). We found that CELSeq produced higher correlations for ES cells and distinguished between cell types more clearly, when examining the highly expressed genes in either cell type (Figures 2A and S2B). Furthermore, CEL-Seq detected significantly more genes in the ES cells
+
+(Figure 2B). In order to ensure a fair comparison between the two methods, we quantified reproducibility for each method separately, using the same criteria as previously described by Islam et al. (2011), and found that with CEL-Seq, significantly lower noise was detected across biological replicates for both cell types tested (Figures 2C and S2C). Finally, principal component analysis based on CEL-Seq data better distinguishes between cell types than the corresponding STRT data (Figure S2D). Biological variation between replicates is a confounding factor when trying to establish the performance of a method. Wetherefore also compared the two methods based on expression levels of exogenously introduced RNA (see below) and found that CEL-Seq provided more reproducible measurements (Figure S2E).
+
+## CEL-Seq Is Highly Sensitive and Reproducible
+
+In order to determine CEL-Seq's sensitivity and reproducibility, we analyzed different amounts of purified C. elegans RNA from mixed embryonic stages, eliminating biological variability present between single cells and allowing us to use different dilutions of the same RNA. We prepared stepwise dilutions, from 40 pg of total RNA down to levels representative of mammalian single cells (  5 pg). In parallel we sequenced a 1 ng RNA sample from the same preparation for use as a reference. In half of the samples, we added exogenous ''carrier'' RNA to test whether the overall amount of RNA in a reverse-transcription reaction affects the efficiency of the reverse-transcription step and found that it did not: the number of reads that mapped to the C. elegans genome depended only on the amount of C. elegans RNA present in a given sample (Figure S3A). Each sample also contained a set of 92 spike-in RNAs with defined concentrations, spanning more than five orders of magnitude (Baker et al., 2005), which showed a linear response across the entire detection range (R 2 = 0.87 ± 0.04 for the 10 pg samples; Figures 3A and S3B).
+
+<!-- image -->
+
+## Figure 3. Sensitivity and Reproducibility of CEL-Seq
+
+(A) CEL-Seq achieves a linear response over the entire detection range. The plot indicates the 92 ERCC (Baker et al., 2005) spike-in levels from six 10 pg replicate samples. Averages for each of the 17 groups of spike-ins with the same nominal concentrationareshownaslargercircles.Foreach sample the spike-ins were normalized such that the average expression of the 12 th spike-in group containing 1,000 molecules was set to 1,000. The fraction of spike-ins in each group without detected expression is shown at the bottom of the figure. The line indicates an idealized linear relationship. Systematic deviations from the line for some spike-ins with high concentrations are likely due to differences in G/C content.
+
+(B) C. elegans transcript counts per sample based on linear regression of spike-in expression levels (5 pg, n = 5; 10 pg, n = 6; 20 pg, n = 5; 40 pg, n = 3). The number of molecules calculated is directly proportional to the amount of input RNA. Error bars indicate the SD.
+
+(C) CEL-Seq sensitivity. For different amounts of input RNA, the bars indicate the percentage of genes detected (at least one read) as a function of their absolute copy number, calculated based on the 1 ng reference sample. Error bars indicate the SDs.
+
+See Figure S3 for additional analyses.
+
+replicates of 20 pg C. elegans RNA. We simulated different sequencing depths and found that at one million reads, 91% of the genes with an average
+
+Wenext invoked the spike-ins to assess CEL-Seq's reproducibility and sensitivity based upon absolute molecule counts. These calculations indicated that 10 pg of the C. elegans RNA contained approximately 280,000 mRNA molecules (Figure 3B). Using that number, and the relative expression level of each gene in the reference sample, we calculated the expected number of transcripts per gene in each of our dilution series and binned the genes in each sample according to this number. Comparison of the different dilutions showed that CEL-Seq's absolute sensitivity does not decrease with smaller starting amounts of RNA (Figures 3C and S3C). For example 50% of all genes with four to five copies and virtually all genes present in more than 50 copies per cell were detected, independent of the total amount of RNA analyzed (Figure 3C). Similarly, the absolute reproducibility of the method is also not compromised by smaller starting amounts (Figure S3C). In summary, CELSeq's performance in estimating a transcript's abundance depends solely on its absolute copy number in the sample. Thus, whereas the IVT imposes a minimal starting amount of  400 pg total RNA for sufficient yield in a single round, the reverse-transcription reaction does not have this restriction. CEL-Seq thus exploits this difference by pooling reverse-transcription reactions together to arrive at the IVT threshold.
+
+Wealso assessed the required sequencing depth for accurate transcriptomic data using CEL-Seq. We created 12 technical expression &gt;100 tpm had an expression level within 20% of that of the averaged expression and that sequencing deeper did not further improve reproducibility. A total of 250,000 reads produced results that were already sufficient for good quantification and very similar to an order of magnitude higher sequencing depth (Figure S3D).
+
+## Transcriptomic Analysis of Single Cells in the Early C. elegans Embryo Identifies Differential and New Expression
+
+Wenext asked whether CEL-Seq can be used to identify biological differences between closely related cells in the C. elegans embryo. C. elegans embryonic development begins with unequal cleavages producing founder cells-termed blastomeres-some of which are depicted in Figure 4B. The AB and P1 sister blastomeres were examined 10 min after cell division. Examining their transcriptomes, we detected 17 genes with a mean 2-fold difference showing significantly different expression (p &lt; 0.05, t test, FDR-corrected; Figure 4A). The most highly expressed of these is mex-3 , whose differential expression is supported by previous work by Draper et al. (1996). Shuffling the groupings of the triplicates resulted in no differentially expressed genes. Next, we sought to proceed in developmental time, using CEL-Seq to further characterize the early embryonic transcriptome. We collected cells by comparing the germ
+
+<!-- image -->
+
+<!-- image -->
+
+Figure 4. Dissecting the Early C. elegans Embryo with CEL-Seq
+
+<!-- image -->
+
+(A) Differential expression analysis in the two-cell stage blastomeres, AB and P1. A t test was made for the 137 genes for which there was expression &gt;100 tpm and at least 2-fold change between the means of the triplicates. The 17 genes with p &lt; 0.05 (FDR corrected) are shown along with the mean expression (right) and standardized expression of triplicates on the left (mean subtracted and SD divided).
+
+(B) The blastomeres examined in this study are abstracted in the cell lineage. The number of new transcripts is indicated under the vertical arrows. The two-sided arrows indicate the number of differentially expressed genes in the anterior versus the posterior blastomeres, respectively. A list of all differentially expressed and new transcripts is in Tables S2 and S3, respectively.
+
+(C) Gene expression levels (log10 tpm; see color scale on right) for the indicated genes; cell lineage is as in (B).
+
+(D) Classification of the AB and P1 blastomeres. For the indicated number of replicates used in the training data, the performance of the machine-learning classifier (see Experimental Procedures) in assigning blastomere identities is shown as a function of the number of genes included for prediction. See Figure S4 for additional analyses.
+
+lineage to the somatic cells: P1 was allowed to divide to its two daughter cells, P2 and EMS, which were analyzed, and similarly, P2 was allowed to divide to P3 and C, which were then analyzed. We also collected the AB daughter cells. We assayed for differential distribution of transcripts in these additional cells similarly to the AB/P1 analysis (Table S2), as well as asking which genes are newly expressed by comparing with the mother cell (Figure 4B; Table S3). The differential distribution between AB/P1 and EMS/P2 has also been tested by microarray and regular RNA-Seq, in both cases using pooled cells, showing good correlation (Figure S4A). New transcription in a particular cell type was scored when the transcript's concentration exceeded 10 tpm and was &lt;1 tpm in the mother cell (median across replicates). We found that in the four-cell stage, only the EMS cell has considerable expression of new transcripts. Among these are med-1 and pes-10 , known to be newly expressed in this cell (Maduro et al., 2001; Seydoux et al., 1996) (Figure 4C). Interest- ingly, newly expressed EMS genes are enriched for transcription factors (4 out of 17; p &lt; 10  3 , hypergeometric distribution): med-1 and three ccch genesccch-2 , F38C2.7 , and Y116A8C.19 -not previously known to be expressed this early in the embryo.
+
+The C and P3 cells are born at the eight-cell stage. Again, the P-lineage cell has new expression of fewer genes than its somatic sister. The C cell expresses more new genes than EMS, suggesting that additional repressors are removed with the progression of development. Importantly, the C and EMS newly transcribed genes share a highly significant overlap (ten genes; p &lt; 10  27 , hypergeometric distribution; see Figure S4B). Out of the 35 C genes, 6 are transcription factors (three times more than expected; p &lt; 10  3 , hypergeometric distribution). This suggests a somatic program evenly kick started upon divergence from P-lineage transcriptional repression, likely by PIE-1 (Seydoux et al., 1996), and a crucial role for specific transcription factors in early development. Finally, we found early expression of 13 genes in P3 (Table S3); of these, onescrm-4 -is a target of deps-1 , a P granule-associated protein (Spike et al., 2008), suggesting that CEL-Seq is indeed detecting expression in a cell previously thought to be transcriptionally inert (Seydoux et al., 1996).
+
+Finally, we were interested in whether transcriptomic profiles obtained with CEL-Seq can be used to computationally predict the identity of unknown cells, as can be the case when analyzing cells from complex tissues. We reasoned that distinguishing between the closely related C. elegans sister blastomeres would pose a veritable challenge and therefore constitute a suitable test case. In order to achieve additional statistical power, we analyzed AB and P1 blastomeres from six more embryos. We built a classifier that uses a training set of sister blastomeres to choose a specified number of maximally informative genes (Experimental Procedures). We then used this information to predict the likely identities of new pairs of sister blastomeres. Surprisingly, in assessing the performance of our classifier using cross-validation (Figures 4D and S4C), we found that for both the two- and the eight-cell stage sisters, predictions can be made with an average 80%-90% success rate based on data from only three embryos. Using four or five replicates, the success rate increased to 90%-100%. For EMS and P2 the classifier still did far better than guessing, but the lack of high success rates even with five replicates indicated that a systematic bias or outlier effect was present in the data. Still, the good success in highly similar blastomeres underscores the potential of our method for future transcriptomic applications.
+
+## DISCUSSION
+
+Single-cell transcriptomics is poised to revolutionize biological research and medical practice (Tang et al., 2011; Wang and Bodovitz, 2010), allowing for unbiased and comprehensive cell characterization. It is acknowledged, that whenever possible, linear amplification by IVT is preferable to exponential amplification by PCR (Tang et al., 2011). Here, we describe a protocol that combines the power of linear amplification by IVT, with a pooling procedure that allows the efficient analysis of many samples in parallel. We presented evidence that CEL-Seq is a sensitive, accurate, and reproducible single-cell transcriptomics method. We tested CEL-Seq on mammalian cells and nematode embryonic blastomeres, made extensive use of a suite of spike-ins, and established the exact sensitivity and reproducibility of the method using extremely low amounts of purified RNA. Here, we review the advantages and limitations of the method and finally consider CEL-Seq's possible applications.
+
+CEL-Seq's key advantage over other protocols (Islam et al., 2011; Tang et al., 2009) arises from its ability to harness the power of IVT, providing both multiplexing and reproducibility. Bypooling many samples to a single IVT, a single round of amplification is sufficient, and CEL-Seq provides significantly reduced hands-on time both for the amplification and downstream processing, allowing for the preparation of dozens of samples for sequencing within 2-3 days. CEL-Seq makes use of commercially available kits for the amplification and sequencing library preparation; only the bar-coded primers and a few enzymes need to be obtained separately. This makes for the cost-effec- tiveness of CEL-Seq. Furthermore, the protocol is complete from start to end: the amplification, library construction, and downstream bioinformatics analysis are seamlessly connected. The amplification can be done for up to 50 cells (samples) per day by a single person. The bar-coded primers are simple to create and manufacture. Because the barcode is 8 bp in length and can be longer, the number of samples that may have unique barcodes in a given IVT is essentially unlimited. These converge to a single sample ready to interface with library preparation enabling the library preparation for  10 of these, or  500 cells. The library construction kit is a standard Illumina kit that itself provides an additional barcode for each library such that multiple IVTs can be analyzed together on the same sequencing lane. Finally, we provide our analysis pipeline ready for integration within the Galaxy framework, such that expression values can easily be obtained within a matter of hours.
+
+<!-- image -->
+
+In addition to working with single cells, CEL-Seq comes with several other desirable properties, such as strand specificity (&gt;98% of exonic reads come from the sense strand) and barcoding efficiency (&gt;96% of the reads contain barcodes). After amplification, the CEL-Seq protocol selects for the single 3 0 -most fragment of each transcript. In contrast to virtually all other RNA-Seq methods, this greatly simplifies the estimation of expression levels because no normalization by gene length is necessary. Thus, it will be of interest to invoke CEL-Seq whenever RNA amplification is necessary, even when not working with single cells.
+
+We found that CEL-Seq outperformed STRT, a previously introduced PCR-based multiplexed single-cell RNA-Seq method (Islam et al., 2011), in terms of robustness, sensitivity, and reproducibility, and suffered from significantly less technical noise. We note that the two methods examine different ends of the mRNA transcript: 5 0 for STRT, but 3 0 for CEL-Seq. In addition there may have been unavoidable differences in culturing conditions of the cell types analyzed. However, several lines of evidence indicate that it is unlikely that these aspects account for the observed performance differences: (1) expression levels were not compared directly across the two methods, (2) significant differences were found for both cell types, and (3) methods were also compared based on spike-ins, which are not affected by these factors.
+
+The limitations of CEL-Seq fall into the categories of specificity to mRNA, 3 0 bias, and sensitivity to small copy numbers. CELSeq does not detect miRNAs and other nonpolyadenylated transcripts. This can be seen as an advantage because the barcoded transcripts are largely depleted of rRNA (&lt;2%), which increases the efficiency of the sequenced reads to measure mRNA levels. Due to its strong 3 0 bias, the method is severely limited in its ability to distinguish alternative splice forms. Another aspect of the 3 0 localization of the reads is that in species with genomes that are not well annotated, the reads will map to unannotated 3 0 UTRs. This could be remedied to some extent by artificially extending transcript annotations beyond the annotated 3 0 end. Finally, a crucial issue with any single-cell gene expression method is the sensitivity to the detection of lowly expressed genes. We have calculated that if the transcript is at five copies, there is a 50% chance of its identification by CEL-Seq. Relative to RNA-Seq of pooled samples, this may seem less sensitive; however, pooling effectively increases the copy number that we have shown to be the important parameter for detection. Nevertheless, CEL-Seq on single cells will capture variation in the expression levels among cells.
+
+<!-- image -->
+
+## EXPERIMENTAL PROCEDURES
+
+## Single-Cell Isolation
+
+C. elegans blastomeres were isolated as previously described by Edgar (1995). Mammaliancells were obtained by trypsin treatment of adherent cells; see also the Extended Experimental Procedures. Individual cells (or media without a cell for negative control) were transferred with a micropipette into a 0.5 m l drop of egg salts or PBS for C. elegans blastomeres or mammalian cells, respectively, placed on the cap of a 0.5 ml LoBind Eppendorf tube, excess liquid was aspirated off, and frozen in liquid nitrogen. Samples were stored at  80  C.
+
+## CEL-Seq Primer Design
+
+The reverse-transcription primer was designed with an anchored polyT, a unique barcode, the 5 0 Illumina adaptor, and a T7 promoter. The T7 promoter sequence was as previously described by Baugh et al. (2001). The Illumina 5 0 adaptor sequence was as used in the Illumina small RNA kit. The barcodes were of length eight and designed in groups of four, such that the first five nucleotides will have equal representation of all four nucleotides to allow for template generation and crosstalk corrections that are based on the first four nucleotides read in the Illumina platform. The barcodes were designed such that each pair is different by at least two nucleotides, so that a single sequencing error will not produce the wrong barcode. All used primers are described in the detailed Extended Experimental Procedures.
+
+## Linear mRNA Amplification
+
+Ambion's MessageAmp II aRNA Kit (AM1751) was used with the following modifications. The polyT primer was replaced with the CEL-Seq primer. The reverse-transcription reaction was performed at one-tenth volume, with 5 ng of primer per reaction. A total of 0.2 m l of the primer mixed with 1 m l of water or 1 m l of a 1:500,000 dilution of the ERCC spike-in kit (a total of 1.2 m l) was added directly to the lid of the Eppendorf tube where the cell was frozen, and incubated at 70  C for 10 min (with the lid of the thermal cycler heated to 70  C). The sample was spun to the bottom of the tube midincubation. After the second-strand synthesis, samples were pooled and cleaned on a single column before proceeding to the IVT reaction at two-fifths volume for 13 hr. RNA was fragmented (one-fifth volume of 200 mM Tris-acetate [pH 8.1], 500 mM KOAc, 150 mM MgOAc added) for 3 min at 94  C, and the reaction was stopped by placing on ice and the addition of one-tenth volume of 0.5 M EDTA, followed by RNA cleanup. The RNA quality and yield were assayed using a Bioanalyzer (Agilent).
+
+## Library Construction and Sequencing
+
+Illumina's directional RNA sequencing protocol was used with the following modifications. A total of 5 ng of RNA was used as input. The mRNA pulldown and fragmentation steps were skipped because amplified RNA represents only mRNA sequences and was already fragmented. Only the 3 0 Illumina adaptor was ligated-diluted 1:5 prior to ligation to obtain the appropriate molar ratio with the reduced amount of RNA. A total of 12 cycles of PCR was performed with an elongation time of 30 s. Libraries were sequenced on the Illumina HiSeq2000 according to standard protocols. Paired-end sequencing was performed, reading at least 15 bases for read 1, and 50 bases for read 2, and the Illumina barcode when needed.
+
+## Expression Analysis Pipeline
+
+Transcript abundances were obtained from the sequencing data using custom scripts organized into a multistep, paralleled computational pipeline within the Galaxy framework (Giardine et al., 2005). Briefly, after trimming and filtering, the paired-end reads were demultiplexed based on the first eight bases of the first read. For each sample, reads were mapped to the C. elegans reference genome (WS230; http://www.wormbase.org), counted using htseq-count (http://www-huber.embl.de/users/anders/HTSeq), and normalized by dividing by the total number of counted reads and multiplying with 10 6 . Because CELSeq retains one fragment per transcript, this procedure yields the estimated gene expression levels in tpm. Absolute copy numbers were obtained by first performing least-squares linear regression on the spike-in values. The resulting factor was used to convert the tpm values to mRNA copy numbers (scripts available at yanailab.technion.ac.il).
+
+## Classification of Blastomere Identities
+
+A machine-learning classifier was devised to predict the identities of pairs of sister blastomeres, and its performance was assessed using cross-validation. Briefly, for a given number of samples to be used as training data, and a given number of genes to be used in the prediction ( G ), the classifier first ranks genes according to the significance of their differential expression in the training data (using a paired t test). The G -most different genes are selected, and their mean differences between the classes are normalized by their SD, to yield a reference vector. For a new set of sister blastomeres, the score for the two possible classifications is calculated as the Euclidean distance between the differences normalized by the SD from the training data, and the reference vector. The predicted classification is chosen according to the smaller distance. The number of possible combinations of data sets to choose for the training step is N choose k, where k is the number of embryos to be used, and N is the total number of embryos analyzed. For each k, all such possibilities are tested, and the average success rate is reported.
+
+## ACCESSION NUMBERS
+
+The NCBI SRA accession number for the sequence data reported in this paper is SRP014672.
+
+## SUPPLEMENTAL INFORMATION
+
+Supplemental Information includes Extended Experimental Procedures, four figures, three tables, and detailed protocol of the CEL-Seq method and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2012. 08.003.
+
+## LICENSING INFORMATION
+
+This is an open-access article distributed under the terms of the Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 Unported License (CC-BY-NC-ND; http://creativecommons.org/licenses/by-nc-nd/3.0/ legalcode).
+
+## ACKNOWLEDGMENTS
+
+This work was supported by the Israel Science Foundation, an FP7 IRG grant and the Lorry I. Lokey Interdisciplinary Center for Life Sciences and Engineering at the Technion. We acknowledge the Gepstein and Aberdam laboratories at the Technion for assistance with the mammalian cells. We also thank Daniel Glikman for a critical reading and advice. T.H., N.S., and I.Y. conceived the method. N.S. led the development of the method. T.H. isolated the blastomeres and mouse cells, performed the validation with other methods, and managed the DNA sequencing. F.W. developed the pipeline to derive the gene expression levels and performed the quality controls. T.H., F.W., and I.Y. analyzed the data. All authors contributed to the experimental designs and the writing of the manuscript.
+
+Received: June 12, 2012
+
+Revised: July 18, 2012
+
+Accepted: August 3, 2012
+
+Published online: August 30, 2012
+
+## REFERENCES
+
+Baker, S.C., Bauer, S.R., Beyer, R.P., Brenton, J.D., Bromley, B., Burrill, J., Causton, H., Conley, M.P., Elespuru, R., Fero, M., et al; External RNA Controls
+
+Consortium. (2005). The External RNA Controls Consortium: a progress report. Nat. Methods 2 , 731-734.
+
+Baugh, L.R., Hill, A.A., Brown, E.L., and Hunter, C.P. (2001). Quantitative analysis of mRNA amplification by in vitro transcription. Nucleic Acids Res. 29 , E29.
+
+Draper, B.W., Mello, C.C., Bowerman, B., Hardin, J., and Priess, J.R. (1996). MEX-3 is a KH domain protein that regulates blastomere identity in early C . elegans embryos. Cell 87 , 205-216.
+
+Eberwine, J., Yeh, H., Miyashiro, K., Cao, Y., Nair, S., Finnell, R., Zettel, M., and Coleman, P. (1992). Analysis of gene expression in single live neurons. Proc. Natl. Acad. Sci. USA 89 , 3010-3014.
+
+Edgar, L.G. (1995). Blastomere culture and analysis. Methods Cell Biol. 48 , 303-321.
+
+Giardine, B., Riemer, C., Hardison, R.C., Burhans, R., Elnitski, L., Shah, P., Zhang, Y., Blankenberg, D., Albert, I., Taylor, J., et al. (2005). Galaxy: a platform for interactive large-scale genome analysis. Genome Res. 15 , 1451-1455.
+
+Islam, S., Kja ¨ llquist, U., Moliner, A., Zajac, P., Fan, J.B., Lo ¨ nnerberg, P., and Linnarsson, S. (2011). Characterization of the single-cell transcriptional landscape by highly multiplex RNA-seq. Genome Res. 21 , 1160-1167.
+
+<!-- image -->
+
+Maduro, M.F., Meneghini, M.D., Bowerman, B., Broitman-Maduro, G., and Rothman, J.H. (2001). Restriction of mesendoderm to a single blastomere by the combined action of SKN-1 and a GSK-3beta homolog is mediated by MED-1 and -2 in C . elegans . Mol. Cell 7 , 475-485.
+
+Seydoux, G., Mello, C.C., Pettitt, J., Wood, W.B., Priess, J.R., and Fire, A. (1996). Repression of gene expression in the embryonic germ lineage of C . elegans . Nature 382 , 713-716.
+
+Spike, C.A., Bader, J., Reinke, V., and Strome, S. (2008). DEPS-1 promotes P-granule assembly and RNA interference in C . elegans germ cells. Development 135 , 983-993.
+
+Tang, F., Barbacioru, C., Wang, Y., Nordman, E., Lee, C., Xu, N., Wang, X., Bodeau, J., Tuch, B.B., Siddiqui, A., et al. (2009). mRNA-Seq whole-transcriptome analysis of a single cell. Nat. Methods 6 , 377-382.
+
+Tang, F., Lao, K., and Surani, M.A. (2011). Development and applications of single-cell transcriptome analysis. Nat. Methods 8 (4, Suppl), S6-S11.
+
+Wang, D., and Bodovitz, S. (2010). Single cell analysis: the new frontier in 'omics'. Trends Biotechnol. 28 , 281-290.
+
+Wang, Z., Gerstein, M., and Snyder, M. (2009). RNA-Seq: a revolutionary tool for transcriptomics. Nat. Rev. Genet. 10 , 57-63.
+
+## Docling layout prediction text cells
+
+### Page 1
+
+Cell Reports
+Resource
+CEL-Seq: Single-Cell RNA-Seq
+by Multiplexed Linear Amplification
+Tamar Hashimshony, 1,2 Florian Wagner, 1,2 Noa Sher, 1,2 and Itai Yanai 1, *
+1 Department of Biology, Technion-Israel Institute of Technology, Haifa 32000, Israel
+2 These authors contributed equally to this work
+*Correspondence: yanai@technion.ac.il
+http://dx.doi.org/10.1016/j.celrep.2012.08.003
+SUMMARY With the dramatically decreasing costs of sequencing, RNA-
+Seq (Wang et al., 2009) has emerged as the preferred method
+High-throughput sequencing has allowed for unprec- for transcriptomic analyses, overtaking microarrays, providing
+edented detail in gene expression analyses, yet its an imperative for any transcriptomic method to be adapted for
+efficient application to single cells is challenged by RNA-Seq. A PCR-based amplification protocol has been used
+the small starting amounts of RNA. We have devel- in combination with SOLID sequencing (Tang et al., 2009).
+oped CEL-Seq, a method for overcoming this limita- Recently, the PCR-based method has been extended to include
+a multiplexing step for the amplification of multiple cells in
+tion by barcoding and pooling samples before
+parallel, allowing for high-throughput analysis, and uses the Illu-
+linearly amplifying mRNA with the use of one round
+mina sequencing platform (Islam et al., 2011). In comparison, IVT
+of in vitro transcription. We show that CEL-Seq gives
+of single cells has been described before; however, it is labor
+more reproducible, linear, and sensitive results than
+intensive (Eberwine et al., 1992), requiring three rounds of
+a PCR-based amplification method. We demon- amplification (  5 days work/cell) and has not been adapted for
+strate the power of this method by studying early multiplexed sequencing. These considerations have hitherto
+embryonic development at single-cell prevented the higher quality possible with IVT from being adapt-
+C. elegans
+resolution. Differential distribution of transcripts ed for single-cell RNA-Seq.
+between sister cells is seen as early as the two-cell Here, we present CEL-Seq (Cell Expression by Linear amplifi-
+stage embryo, and zygotic expression in the somatic cation and Sequencing), a protocol that meets the demand of
+cell lineages is enriched for transcription factors. The linear amplification by IVT for sufficient material by pooling bar-
+coded samples, therefore allowing the efficient linear amplifica-
+robust transcriptome quantifications enabled by
+tion of RNA from single cells and their analysis by sequencing.
+CEL-Seq will be useful for transcriptomic analyses
+Wecompare the performance of our method on two mammalian
+of complex tissues containing populations of diverse
+cell types to that of a PCR-based approach and use spike-ins to
+cell types.
+establish CEL-Seq's exact reproducibility and sensitivity at very
+low amounts of input RNA. Finally, we apply our protocol to study
+sister cells from early C. elegans embryos, and demonstrate that
+INTRODUCTION
+CEL-Seq's high performance can be used to reliably distinguish
+between cell types, even in cases where only subtle biological
+For many biological questions a single-cell-level description of
+differences are present.
+gene regulation is advantageous to cell populations (Tang
+et al., 2011; Wang and Bodovitz, 2010). Microscopy, FACS, or
+real-time PCR-based methods can provide a single-cell aspect RESULTS
+to experiments but are able to assay only a handful of genes at
+a time. High-throughput technologies such as microarrays and CEL-Seq Performs Multiplexed Single-Cell
+RNA-Seq provide a full view of the expression of all genes but Transcriptomics by Linear Amplification
+are limited by the amount of RNA needed for analysis. This can The CEL-Seq method begins with a single-cell reverse-tran-
+be solved by adding an RNA amplification step, either by expo- scription reaction using a primer designed with an anchored
+nential PCR-based amplification or linear in vitro transcription polyT, a unique barcode, the 5 0 Illumina sequencing adaptor,
+(IVT) amplification (Eberwine et al., 1992). With PCR practically and a T7 promoter (Figure 1A; see Experimental Procedures
+any RNA starting amount can be employed, simply by adding for details). Next, second-strand synthesis is performed and
+additional cycles, thereby allowing analysis at the single-cell then the cDNA samples are pooled and consequently comprise
+level. However, efforts for linear amplification of RNA from single sufficient template material for an IVT reaction. The amplified
+cells have been challenged by IVT's lower bound of  400 pg RNA is then subjected to directional RNA library preparation.
+total RNA as input material for a single round of amplification. The RNA is fragmented to a size distribution appropriate for
+Therefore, to date, IVT has not been efficiently used for amplifi- sequencing, the Illumina 3 0 adaptor is added by ligation, RNA
+cation of RNA from single cells (Tang et al., 2011). is reverse transcribed to DNA, and the 3 0 -most fragments that
+666 Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors
+
+### Page 2
+
+A
+C E
+B
+D
+Figure 1. The CEL-Seq Method
+(A) Individual cells are added to tubes, each with a uniquely bar-coded primer for reverse transcription. After second-strand synthesis, the reactions are pooled for
+IVT. The amplified RNA is then fragmented and purified before entry into a modified version of the Illumina directional RNA protocol, the molecules with both
+Illumina adaptors are selected, and the DNA library is sequenced with paired-end reads.
+(B) Nucleotide distribution in the sequenced paired-end reads. Each nucleotide position is represented by one column, with the first base on the left.
+(C) Barcode distribution of one IVT reaction after demultiplexing. The cells from three two-cell stage C. elegans embryos (denoted P1 and AB) and a single one-cell
+stage embryo (denoted P0) were amplified together in a single multiplexed IVT reaction.
+(D) Distribution of the reads mapping to the C. elegans genome in the six AB/P1 cells. Error bars indicate the SD.
+(E) Correlation between biological AB replicates.
+See also Figure S1C.
+contain both Illumina adaptors and a barcode are selected. The precisely at the beginning of the first read, invariably followed
+resulting library undergoes paired-end sequencing, where the by a polyT stretch (Figure 1B). Barcodes from all six samples
+first read recovers the barcode, whereas the second identifies were represented, indicating the success of the individual
+the mRNA transcript (Figure 1A). Thus, by multiplexing CEL- single-cell reverse-transcription reactions (Figure 1C). We map-
+Seq takes advantage of the different input requirements of the ped reads to the C. elegans genome and found that 91.7%
+reverse-transcription and IVT reactions to obtain sufficient RNA stemmed from mRNA. Only 2.0% stemmed from ribosomal
+from single cells for a single round of linear amplification. RNA(rRNA), demonstrating CEL-Seq's specificity for polyadeny-
+As an initial test, we applied CEL-Seq to individual cells iso- lated transcripts. This was also supported by the extremely low-
+lated from three two-cell C. elegans embryos (Table S1). On detected expression levels of core histone mRNAs, which are
+average, 95.5% of the filtered reads had a barcode located thought to be highly expressed yet mostly nonpolyadenylated
+Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors 667
+
+### Page 3
+
+A B C
+Figure 2. Benchmarking of CEL-Seq on Mouse ES and MEF Cells
+(A) CEL-Seq Pearson's correlation coefficients among the ES and MEF cells (on log10 tpm values), computed as previously described by Islam et al. (2011), on the
+1,000 genes most highly expressed in ES cells and 1,000 genes most highly expressed in MEF cells (1,385 genes).
+(B) Mean number of genes detected above two thresholds (10 and 100 tpm) across the ES and MEF cell types using CEL-Seq and the ''STRT'' PCR-based
+method (Islam et al., 2011). Error bars indicate 95% confidence intervals.
+(C) Reproducibility according to expression level. For each gene the coefficient of variation was computed across the log10 tpm values in the ES cells for the STRT
+and CEL-Seq methods. The genes were then ranked by expression level in bins of 200 from high to low. For each bin the mean and SD of the coefficients of
+variation of the genes are shown.
+See also Figure S2 for additional analyses.
+(Figure S1A). Finally, RNase treatment of cells did not (Figure 2B). In order to ensure a fair comparison between the
+produce amplified RNA indicating the specificity of the method two methods, we quantified reproducibility for each method
+to RNA. separately, using the same criteria as previously described by
+CEL-Seq is highly strand specific because 97.2% of exonic Islam et al. (2011), and found that with CEL-Seq, significantly
+reads exhibited sense orientation (Figure 1D). The reads mapped lower noise was detected across biological replicates for both
+exclusively to the 3 0 end of transcripts (Figure S1B), which was cell types tested (Figures 2C and S2C). Finally, principal compo-
+expected because CEL-Seq only retains the 3 0 -most fragments nent analysis based on CEL-Seq data better distinguishes
+of transcripts (Figure 1A). Some reads mapped to intergenic between cell types than the corresponding STRT data (Fig-
+sequences, but manual inspection of the aligned reads revealed ure S2D). Biological variation between replicates is a confound-
+that this can likely be explained by incomplete 3 0 UTR annota- ing factor when trying to establish the performance of a method.
+tions (data not shown). Expression levels were then estimated Wetherefore also compared the two methods based on expres-
+by counting all reads mapping to each gene, and normalized to sion levels of exogenously introduced RNA (see below) and
+give the read count in transcripts per million (tpm; see Experi- found that CEL-Seq provided more reproducible measurements
+mental Procedures). The expression levels of all genes (hence- (Figure S2E).
+forth, transcriptome) across biological replicates showed an
+average correlation of R = 0.979 (Figures 1E and S1C). A nega- CEL-Seq Is Highly Sensitive and Reproducible
+tive control-starting with the growth media lacking a cell- In order to determine CEL-Seq's sensitivity and reproducibility,
+resulted in very few reads (Figure S1D). we analyzed different amounts of purified C. elegans RNA from
+mixed embryonic stages, eliminating biological variability
+CEL-Seq Outperforms a PCR-Based Multiplexed RNA- present between single cells and allowing us to use different dilu-
+Seq Method tions of the same RNA. We prepared stepwise dilutions, from
+We next sought to compare the performance of our protocol to 40 pg of total RNA down to levels representative of mammalian
+the STRT method, a previously introduced PCR-based multi- single cells (  5 pg). In parallel we sequenced a 1 ng RNA sample
+plexed RNA-Seq method by Islam et al. (2011). We thus applied from the same preparation for use as a reference. In half of the
+CEL-Seq to the cell types compared in this previous study by samples, we added exogenous ''carrier'' RNA to test whether
+Islam et al. (2011) and determined the transcriptomes of nine the overall amount of RNA in a reverse-transcription reaction
+mouse embryonic stem (ES) cells and seven mouse embryonic affects the efficiency of the reverse-transcription step and found
+fibroblasts (MEFs) (Table S1). When comparing the distribution that it did not: the number of reads that mapped to the C. elegans
+of expression levels of each single-cell transcriptome across genome depended only on the amount of C. elegans RNA
+methods and cell type, CEL-Seq shows more reproducible present in a given sample (Figure S3A). Each sample also con-
+distributions of expression (Figure S2A). We found that CEL- tained a set of 92 spike-in RNAs with defined concentrations,
+Seq produced higher correlations for ES cells and distinguished spanning more than five orders of magnitude (Baker et al.,
+between cell types more clearly, when examining the highly ex- 2005), which showed a linear response across the entire detec-
+pressed genes in either cell type (Figures 2A and S2B). Further- tion range (R 2 = 0.87 ± 0.04 for the 10 pg samples; Figures 3A
+more, CEL-Seq detected significantly more genes in the ES cells and S3B).
+668 Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors
+
+### Page 4
+
+A B Figure 3. Sensitivity and Reproducibility of
+CEL-Seq
+(A) CEL-Seq achieves a linear response over the
+entire detection range. The plot indicates the 92
+ERCC (Baker et al., 2005) spike-in levels from six
+10 pg replicate samples. Averages for each of the
+17 groups of spike-ins with the same nominal
+concentrationareshownaslargercircles.Foreach
+sample the spike-ins were normalized such that
+the average expression of the 12 th spike-in group
+containing 1,000 molecules was set to 1,000. The
+fraction of spike-ins in each group without de-
+tected expression is shown at the bottom of the
+figure. The line indicates an idealized linear rela-
+tionship. Systematic deviations from the line for
+C some spike-ins with high concentrations are likely
+due to differences in G/C content.
+(B) C. elegans transcript counts per sample based
+on linear regression of spike-in expression levels
+(5 pg, n = 5; 10 pg, n = 6; 20 pg, n = 5; 40 pg, n = 3).
+The number of molecules calculated is directly
+proportional to the amount of input RNA. Error bars
+indicate the SD.
+(C) CEL-Seq sensitivity. For different amounts of
+input RNA, the bars indicate the percentage of
+genes detected (at least one read) as a function
+of their absolute copy number, calculated based
+on the 1 ng reference sample. Error bars indicate
+the SDs.
+See Figure S3 for additional analyses.
+replicates of 20 pg C. elegans RNA. We
+simulated different sequencing depths
+and found that at one million reads,
+91% of the genes with an average
+Wenext invoked the spike-ins to assess CEL-Seq's reproduc- expression >100 tpm had an expression level within 20% of
+ibility and sensitivity based upon absolute molecule counts. that of the averaged expression and that sequencing deeper
+These calculations indicated that 10 pg of the C. elegans RNA did not further improve reproducibility. A total of 250,000 reads
+contained approximately 280,000 mRNA molecules (Figure 3B). produced results that were already sufficient for good quantifica-
+Using that number, and the relative expression level of each tion and very similar to an order of magnitude higher sequencing
+gene in the reference sample, we calculated the expected depth (Figure S3D).
+number of transcripts per gene in each of our dilution series
+and binned the genes in each sample according to this number. Transcriptomic Analysis of Single Cells in the Early
+Comparison of the different dilutions showed that CEL-Seq's Embryo Identifies Differential and New
+C. elegans
+absolute sensitivity does not decrease with smaller starting Expression
+amounts of RNA (Figures 3C and S3C). For example 50% of all Wenext asked whether CEL-Seq can be used to identify biolog-
+genes with four to five copies and virtually all genes present in ical differences between closely related cells in the C. elegans
+more than 50 copies per cell were detected, independent of embryo. C. elegans embryonic development begins with
+the total amount of RNA analyzed (Figure 3C). Similarly, the unequal cleavages producing founder cells-termed blasto-
+absolute reproducibility of the method is also not compromised meres-some of which are depicted in Figure 4B. The AB and
+by smaller starting amounts (Figure S3C). In summary, CEL- P1 sister blastomeres were examined 10 min after cell division.
+Seq's performance in estimating a transcript's abundance Examining their transcriptomes, we detected 17 genes with
+depends solely on its absolute copy number in the sample. a mean 2-fold difference showing significantly different expres-
+Thus, whereas the IVT imposes a minimal starting amount of sion (p < 0.05, t test, FDR-corrected; Figure 4A). The most highly
+ 400 pg total RNA for sufficient yield in a single round, the expressed of these is mex-3 , whose differential expression is
+reverse-transcription reaction does not have this restriction. supported by previous work by Draper et al. (1996). Shuffling
+CEL-Seq thus exploits this difference by pooling reverse-tran- the groupings of the triplicates resulted in no differentially ex-
+scription reactions together to arrive at the IVT threshold. pressed genes. Next, we sought to proceed in developmental
+Wealso assessed the required sequencing depth for accurate time, using CEL-Seq to further characterize the early embryonic
+transcriptomic data using CEL-Seq. We created 12 technical transcriptome. We collected cells by comparing the germ
+Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors 669
+
+### Page 5
+
+A B
+C D
+Figure 4. Dissecting the Early Embryo with CEL-Seq
+C. elegans
+(A) Differential expression analysis in the two-cell stage blastomeres, AB and P1. A t test was made for the 137 genes for which there was expression >100 tpm
+and at least 2-fold change between the means of the triplicates. The 17 genes with p < 0.05 (FDR corrected) are shown along with the mean expression (right) and
+standardized expression of triplicates on the left (mean subtracted and SD divided).
+(B) The blastomeres examined in this study are abstracted in the cell lineage. The number of new transcripts is indicated under the vertical arrows. The two-sided
+arrows indicate the number of differentially expressed genes in the anterior versus the posterior blastomeres, respectively. A list of all differentially expressed and
+new transcripts is in Tables S2 and S3, respectively.
+(C) Gene expression levels (log10 tpm; see color scale on right) for the indicated genes; cell lineage is as in (B).
+(D) Classification of the AB and P1 blastomeres. For the indicated number of replicates used in the training data, the performance of the machine-learning
+classifier (see Experimental Procedures) in assigning blastomere identities is shown as a function of the number of genes included for prediction.
+See Figure S4 for additional analyses.
+lineage to the somatic cells: P1 was allowed to divide to its two ingly, newly expressed EMS genes are enriched for transcription
+daughter cells, P2 and EMS, which were analyzed, and similarly, factors (4 out of 17; p < 10  3 , hypergeometric distribution): med-1
+P2 was allowed to divide to P3 and C, which were then analyzed. and three ccch genes- ccch-2 , F38C2.7 , and Y116A8C.19 -not
+We also collected the AB daughter cells. We assayed for differ- previously known to be expressed this early in the embryo.
+ential distribution of transcripts in these additional cells similarly The C and P3 cells are born at the eight-cell stage. Again, the
+to the AB/P1 analysis (Table S2), as well as asking which genes P-lineage cell has new expression of fewer genes than its
+are newly expressed by comparing with the mother cell (Fig- somatic sister. The C cell expresses more new genes than
+ure 4B; Table S3). The differential distribution between AB/P1 EMS, suggesting that additional repressors are removed with
+and EMS/P2 has also been tested by microarray and regular the progression of development. Importantly, the C and EMS
+RNA-Seq, in both cases using pooled cells, showing good corre- newly transcribed genes share a highly significant overlap (ten
+lation (Figure S4A). New transcription in a particular cell type was genes; p < 10  27 , hypergeometric distribution; see Figure S4B).
+scored when the transcript's concentration exceeded 10 tpm Out of the 35 C genes, 6 are transcription factors (three times
+and was <1 tpm in the mother cell (median across replicates). more than expected; p < 10  3 , hypergeometric distribution).
+We found that in the four-cell stage, only the EMS cell has This suggests a somatic program evenly kick started upon diver-
+considerable expression of new transcripts. Among these are gence from P-lineage transcriptional repression, likely by PIE-1
+med-1 and pes-10 , known to be newly expressed in this cell (Seydoux et al., 1996), and a crucial role for specific transcription
+(Maduro et al., 2001; Seydoux et al., 1996) (Figure 4C). Interest- factors in early development. Finally, we found early expression
+670 Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors
+
+### Page 6
+
+of 13 genes in P3 (Table S3); of these, one- scrm-4 -is a target tiveness of CEL-Seq. Furthermore, the protocol is complete
+of deps-1 , a P granule-associated protein (Spike et al., 2008), from start to end: the amplification, library construction, and
+suggesting that CEL-Seq is indeed detecting expression in downstream bioinformatics analysis are seamlessly connected.
+a cell previously thought to be transcriptionally inert (Seydoux The amplification can be done for up to 50 cells (samples) per
+et al., 1996). day by a single person. The bar-coded primers are simple to
+Finally, we were interested in whether transcriptomic profiles create and manufacture. Because the barcode is 8 bp in length
+obtained with CEL-Seq can be used to computationally predict and can be longer, the number of samples that may have unique
+the identity of unknown cells, as can be the case when analyzing barcodes in a given IVT is essentially unlimited. These converge
+cells from complex tissues. We reasoned that distinguishing to a single sample ready to interface with library preparation
+between the closely related C. elegans sister blastomeres would enabling the library preparation for  10 of these, or  500 cells.
+pose a veritable challenge and therefore constitute a suitable The library construction kit is a standard Illumina kit that itself
+test case. In order to achieve additional statistical power, we provides an additional barcode for each library such that multiple
+analyzed AB and P1 blastomeres from six more embryos. We IVTs can be analyzed together on the same sequencing lane.
+built a classifier that uses a training set of sister blastomeres to Finally, we provide our analysis pipeline ready for integration
+choose a specified number of maximally informative genes within the Galaxy framework, such that expression values can
+(Experimental Procedures). We then used this information to easily be obtained within a matter of hours.
+predict the likely identities of new pairs of sister blastomeres. In addition to working with single cells, CEL-Seq comes with
+Surprisingly, in assessing the performance of our classifier using several other desirable properties, such as strand specificity
+cross-validation (Figures 4D and S4C), we found that for both the (>98% of exonic reads come from the sense strand) and barcod-
+two- and the eight-cell stage sisters, predictions can be made ing efficiency (>96% of the reads contain barcodes). After ampli-
+with an average 80%-90% success rate based on data from fication, the CEL-Seq protocol selects for the single 3 0 -most
+only three embryos. Using four or five replicates, the success fragment of each transcript. In contrast to virtually all other
+rate increased to 90%-100%. For EMS and P2 the classifier still RNA-Seq methods, this greatly simplifies the estimation of
+did far better than guessing, but the lack of high success rates expression levels because no normalization by gene length is
+even with five replicates indicated that a systematic bias or necessary. Thus, it will be of interest to invoke CEL-Seq when-
+outlier effect was present in the data. Still, the good success in ever RNA amplification is necessary, even when not working
+highly similar blastomeres underscores the potential of our with single cells.
+method for future transcriptomic applications. We found that CEL-Seq outperformed STRT, a previously
+introduced PCR-based multiplexed single-cell RNA-Seq
+DISCUSSION method (Islam et al., 2011), in terms of robustness, sensitivity,
+and reproducibility, and suffered from significantly less technical
+Single-cell transcriptomics is poised to revolutionize biological noise. We note that the two methods examine different ends of
+research and medical practice (Tang et al., 2011; Wang and Bod- the mRNA transcript: 5 0 for STRT, but 3 0 for CEL-Seq. In addition
+ovitz, 2010), allowing for unbiased and comprehensive cell char- there may have been unavoidable differences in culturing condi-
+acterization. It is acknowledged, that whenever possible, linear tions of the cell types analyzed. However, several lines of
+amplification by IVT is preferable to exponential amplification evidence indicate that it is unlikely that these aspects account
+by PCR (Tang et al., 2011). Here, we describe a protocol that for the observed performance differences: (1) expression levels
+combines the power of linear amplification by IVT, with a pooling were not compared directly across the two methods, (2) signifi-
+procedure that allows the efficient analysis of many samples in cant differences were found for both cell types, and (3) methods
+parallel. We presented evidence that CEL-Seq is a sensitive, were also compared based on spike-ins, which are not affected
+accurate, and reproducible single-cell transcriptomics method. by these factors.
+We tested CEL-Seq on mammalian cells and nematode embry- The limitations of CEL-Seq fall into the categories of specificity
+onic blastomeres, made extensive use of a suite of spike-ins, and to mRNA, 3 0 bias, and sensitivity to small copy numbers. CEL-
+established the exact sensitivity and reproducibility of the Seq does not detect miRNAs and other nonpolyadenylated tran-
+method using extremely low amounts of purified RNA. Here, scripts. This can be seen as an advantage because the bar-
+we review the advantages and limitations of the method and coded transcripts are largely depleted of rRNA (<2%), which
+finally consider CEL-Seq's possible applications. increases the efficiency of the sequenced reads to measure
+CEL-Seq's key advantage over other protocols (Islam et al., mRNA levels. Due to its strong 3 0 bias, the method is severely
+2011; Tang et al., 2009) arises from its ability to harness the limited in its ability to distinguish alternative splice forms.
+power of IVT, providing both multiplexing and reproducibility. Another aspect of the 3 0 localization of the reads is that in species
+Bypooling many samples to a single IVT, a single round of ampli- with genomes that are not well annotated, the reads will map to
+fication is sufficient, and CEL-Seq provides significantly reduced unannotated 3 0 UTRs. This could be remedied to some extent by
+hands-on time both for the amplification and downstream pro- artificially extending transcript annotations beyond the anno-
+cessing, allowing for the preparation of dozens of samples for tated 3 0 end. Finally, a crucial issue with any single-cell gene
+sequencing within 2-3 days. CEL-Seq makes use of commer- expression method is the sensitivity to the detection of lowly ex-
+cially available kits for the amplification and sequencing library pressed genes. We have calculated that if the transcript is at five
+preparation; only the bar-coded primers and a few enzymes copies, there is a 50% chance of its identification by CEL-Seq.
+need to be obtained separately. This makes for the cost-effec- Relative to RNA-Seq of pooled samples, this may seem less
+Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors 671
+
+### Page 7
+
+sensitive; however, pooling effectively increases the copy by the total number of counted reads and multiplying with 10 6 . Because CEL-
+number that we have shown to be the important parameter for Seq retains one fragment per transcript, this procedure yields the estimated
+gene expression levels in tpm. Absolute copy numbers were obtained by first
+detection. Nevertheless, CEL-Seq on single cells will capture
+performing least-squares linear regression on the spike-in values. The result-
+variation in the expression levels among cells.
+ing factor was used to convert the tpm values to mRNA copy numbers (scripts
+available at yanailab.technion.ac.il).
+EXPERIMENTAL PROCEDURES
+Classification of Blastomere Identities
+Single-Cell Isolation A machine-learning classifier was devised to predict the identities of pairs of
+C. elegans blastomeres were isolated as previously described by Edgar (1995). sister blastomeres, and its performance was assessed using cross-validation.
+Mammaliancells were obtained by trypsin treatment of adherent cells; see also Briefly, for a given number of samples to be used as training data, and a given
+the Extended Experimental Procedures. Individual cells (or media without a cell number of genes to be used in the prediction ( G ), the classifier first ranks genes
+for negative control) were transferred with a micropipette into a 0.5 m l drop of according to the significance of their differential expression in the training data
+egg salts or PBS for C. elegans blastomeres or mammalian cells, respectively, (using a paired t test). The G -most different genes are selected, and their mean
+placed on the cap of a 0.5 ml LoBind Eppendorf tube, excess liquid was aspi- differences between the classes are normalized by their SD, to yield a refer-
+rated off, and frozen in liquid nitrogen. Samples were stored at  80  C. ence vector. For a new set of sister blastomeres, the score for the two possible
+classifications is calculated as the Euclidean distance between the differences
+CEL-Seq Primer Design normalized by the SD from the training data, and the reference vector. The pre-
+The reverse-transcription primer was designed with an anchored polyT, dicted classification is chosen according to the smaller distance. The number
+a unique barcode, the 5 0 Illumina adaptor, and a T7 promoter. The T7 promoter of possible combinations of data sets to choose for the training step is N
+sequence was as previously described by Baugh et al. (2001). The Illumina 5 0 choose k, where k is the number of embryos to be used, and N is the total
+adaptor sequence was as used in the Illumina small RNA kit. The barcodes number of embryos analyzed. For each k, all such possibilities are tested,
+were of length eight and designed in groups of four, such that the first five and the average success rate is reported.
+nucleotides will have equal representation of all four nucleotides to allow for
+template generation and crosstalk corrections that are based on the first
+ACCESSION NUMBERS
+four nucleotides read in the Illumina platform. The barcodes were designed
+such that each pair is different by at least two nucleotides, so that a single
+The NCBI SRA accession number for the sequence data reported in this paper
+sequencing error will not produce the wrong barcode. All used primers are
+is SRP014672.
+described in the detailed Extended Experimental Procedures.
+SUPPLEMENTAL INFORMATION
+Linear mRNA Amplification
+Ambion's MessageAmp II aRNA Kit (AM1751) was used with the following
+Supplemental Information includes Extended Experimental Procedures, four
+modifications. The polyT primer was replaced with the CEL-Seq primer. The
+figures, three tables, and detailed protocol of the CEL-Seq method and can
+reverse-transcription reaction was performed at one-tenth volume, with 5 ng
+be found with this article online at http://dx.doi.org/10.1016/j.celrep.2012.
+of primer per reaction. A total of 0.2 m l of the primer mixed with 1 m l of water
+or 1 m l of a 1:500,000 dilution of the ERCC spike-in kit (a total of 1.2 m l) was 08.003.
+added directly to the lid of the Eppendorf tube where the cell was frozen,
+and incubated at 70  C for 10 min (with the lid of the thermal cycler heated to LICENSING INFORMATION
+70  C). The sample was spun to the bottom of the tube midincubation. After
+the second-strand synthesis, samples were pooled and cleaned on a single This is an open-access article distributed under the terms of the Creative
+column before proceeding to the IVT reaction at two-fifths volume for 13 hr. Commons Attribution-Noncommercial-No Derivative Works 3.0 Unported
+RNA was fragmented (one-fifth volume of 200 mM Tris-acetate [pH 8.1], License (CC-BY-NC-ND; http://creativecommons.org/licenses/by-nc-nd/3.0/
+500 mM KOAc, 150 mM MgOAc added) for 3 min at 94  C, and the reaction legalcode).
+was stopped by placing on ice and the addition of one-tenth volume of
+0.5 M EDTA, followed by RNA cleanup. The RNA quality and yield were as- ACKNOWLEDGMENTS
+sayed using a Bioanalyzer (Agilent).
+This work was supported by the Israel Science Foundation, an FP7 IRG grant
+Library Construction and Sequencing and the Lorry I. Lokey Interdisciplinary Center for Life Sciences and Engi-
+Illumina's directional RNA sequencing protocol was used with the following neering at the Technion. We acknowledge the Gepstein and Aberdam labora-
+modifications. A total of 5 ng of RNA was used as input. The mRNA pull- tories at the Technion for assistance with the mammalian cells. We also thank
+down and fragmentation steps were skipped because amplified RNA repre- Daniel Glikman for a critical reading and advice. T.H., N.S., and I.Y. conceived
+sents only mRNA sequences and was already fragmented. Only the 3 0 Illumina the method. N.S. led the development of the method. T.H. isolated the blasto-
+adaptor was ligated-diluted 1:5 prior to ligation to obtain the appropriate meres and mouse cells, performed the validation with other methods, and
+molar ratio with the reduced amount of RNA. A total of 12 cycles of PCR managed the DNA sequencing. F.W. developed the pipeline to derive the
+was performed with an elongation time of 30 s. Libraries were sequenced on gene expression levels and performed the quality controls. T.H., F.W., and
+the Illumina HiSeq2000 according to standard protocols. Paired-end I.Y. analyzed the data. All authors contributed to the experimental designs
+sequencing was performed, reading at least 15 bases for read 1, and 50 bases and the writing of the manuscript.
+for read 2, and the Illumina barcode when needed.
+Received: June 12, 2012
+Expression Analysis Pipeline Revised: July 18, 2012
+Transcript abundances were obtained from the sequencing data using custom Accepted: August 3, 2012
+scripts organized into a multistep, paralleled computational pipeline within the Published online: August 30, 2012
+Galaxy framework (Giardine et al., 2005). Briefly, after trimming and filtering,
+the paired-end reads were demultiplexed based on the first eight bases of REFERENCES
+the first read. For each sample, reads were mapped to the C. elegans reference
+genome (WS230; http://www.wormbase.org), counted using htseq-count Baker, S.C., Bauer, S.R., Beyer, R.P., Brenton, J.D., Bromley, B., Burrill, J.,
+(http://www-huber.embl.de/users/anders/HTSeq), and normalized by dividing Causton, H., Conley, M.P., Elespuru, R., Fero, M., et al; External RNA Controls
+672 Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors
+
+### Page 8
+
+Consortium. (2005). The External RNA Controls Consortium: a progress report. Maduro, M.F., Meneghini, M.D., Bowerman, B., Broitman-Maduro, G., and
+Nat. Methods 2 , 731-734. Rothman, J.H. (2001). Restriction of mesendoderm to a single blastomere
+by the combined action of SKN-1 and a GSK-3beta homolog is mediated by
+Baugh, L.R., Hill, A.A., Brown, E.L., and Hunter, C.P. (2001). Quantitative anal-
+MED-1 and -2 in C . elegans . Mol. Cell 7 , 475-485.
+ysis of mRNA amplification by in vitro transcription. Nucleic Acids Res. 29 , E29.
+Seydoux, G., Mello, C.C., Pettitt, J., Wood, W.B., Priess, J.R., and Fire, A.
+Draper, B.W., Mello, C.C., Bowerman, B., Hardin, J., and Priess, J.R. (1996).
+(1996). Repression of gene expression in the embryonic germ lineage of
+MEX-3 is a KH domain protein that regulates blastomere identity in early
+C . elegans . Nature 382 , 713-716.
+C . elegans embryos. Cell 87 , 205-216.
+Spike, C.A., Bader, J., Reinke, V., and Strome, S. (2008). DEPS-1 promotes
+Eberwine, J., Yeh, H., Miyashiro, K., Cao, Y., Nair, S., Finnell, R., Zettel, M., and P-granule assembly and RNA interference in C . elegans germ cells. Develop-
+Coleman, P. (1992). Analysis of gene expression in single live neurons. Proc. ment 135 , 983-993.
+Natl. Acad. Sci. USA 89 , 3010-3014.
+Tang, F., Barbacioru, C., Wang, Y., Nordman, E., Lee, C., Xu, N., Wang, X.,
+Edgar, L.G. (1995). Blastomere culture and analysis. Methods Cell Biol. 48 , Bodeau, J., Tuch, B.B., Siddiqui, A., et al. (2009). mRNA-Seq whole-transcrip-
+303-321. tome analysis of a single cell. Nat. Methods 6 , 377-382.
+Giardine, B., Riemer, C., Hardison, R.C., Burhans, R., Elnitski, L., Shah, P., Tang, F., Lao, K., and Surani, M.A. (2011). Development and applications of
+Zhang, Y., Blankenberg, D., Albert, I., Taylor, J., et al. (2005). Galaxy: a platform single-cell transcriptome analysis. Nat. Methods 8 (4, Suppl), S6-S11.
+for interactive large-scale genome analysis. Genome Res. 15 , 1451-1455. Wang, D., and Bodovitz, S. (2010). Single cell analysis: the new frontier in
+Islam, S., Kja ¨ llquist, U., Moliner, A., Zajac, P., Fan, J.B., Lo ¨ nnerberg, P., and 'omics'. Trends Biotechnol. 28 , 281-290.
+Linnarsson, S. (2011). Characterization of the single-cell transcriptional land- Wang, Z., Gerstein, M., and Snyder, M. (2009). RNA-Seq: a revolutionary tool
+scape by highly multiplex RNA-seq. Genome Res. 21 , 1160-1167. for transcriptomics. Nat. Rev. Genet. 10 , 57-63.
+Cell Reports 2 , 666-673, September 27, 2012 ª 2012 The Authors 673

cel_seq/CEL-Seq_paper.pymupdf_text.txt

@@ -0,0 +1,483 @@
+# pymupdf text extraction
+source_pdf: CEL-Seq_paper.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/cel_seq/CEL-Seq_paper.pdf
+extraction: PyMuPDF: page.get_text("text", sort=True)
+
+## Page 1
+
+                                                                                         Cell Reports
+                          Resource
+
+
+CEL-Seq: Single-Cell RNA-Seq
+by Multiplexed Linear Amplification
+
+Tamar Hashimshony,1,2 Florian Wagner,1,2 Noa Sher,1,2 and Itai Yanai1,*
+1Department of Biology, Technion-Israel Institute of Technology, Haifa 32000, Israel
+2These authors contributed equally to this work
+*Correspondence: yanai@technion.ac.il
+ http://dx.doi.org/10.1016/j.celrep.2012.08.003
+
+
+
+
+SUMMARY                                                  With the dramatically decreasing costs of sequencing, RNA-
+                                                  Seq (Wang et al., 2009) has emerged as the preferred method
+High-throughput sequencing has allowed for unprec-    for transcriptomic analyses, overtaking microarrays, providing
+edented detail in gene expression analyses, yet its   an imperative for any transcriptomic method to be adapted for
+efficient application to single cells is challenged by   RNA-Seq. A PCR-based amplification protocol has been used
+the small starting amounts of RNA. We have devel-    in combination with SOLID sequencing (Tang et  al., 2009).
+oped CEL-Seq, a method for overcoming this limita-   Recently, the PCR-based method has been extended to include
+                                                       a multiplexing step for the amplification of multiple cells in
+tion by barcoding and pooling samples before
+                                                                             parallel, allowing for high-throughput analysis, and uses the Illu-
+linearly amplifying mRNA with the use of one round
+                                                       mina sequencing platform (Islam et al., 2011). In comparison, IVT
+of in vitro transcription. We show that CEL-Seq gives                                                                     of single cells has been described before; however, it is labor
+more reproducible, linear, and sensitive results than   intensive (Eberwine et  al., 1992), requiring three rounds of
+a PCR-based amplification method. We demon-   amplification ( 5 days work/cell) and has not been adapted for
+strate the power of this method by studying early   multiplexed sequencing. These considerations have hitherto
+C. elegans embryonic development at single-cell   prevented the higher quality possible with IVT from being adapt-
+resolution.  Differential  distribution  of  transcripts   ed for single-cell RNA-Seq.
+between sister cells is seen as early as the two-cell     Here, we present CEL-Seq (Cell Expression by Linear amplifi-
+stage embryo, and zygotic expression in the somatic   cation and Sequencing), a protocol that meets the demand of
+cell lineages is enriched for transcription factors. The    linear amplification by IVT for sufficient material by pooling bar-
+                                                    coded samples, therefore allowing the efficient linear amplifica-
+robust transcriptome quantifications enabled by
+                                                                        tion of RNA from single cells and their analysis by sequencing.
+CEL-Seq will be useful for transcriptomic analyses
+                                        We compare the performance of our method on two mammalian
+of complex tissues containing populations of diverse                                                                              cell types to that of a PCR-based approach and use spike-ins to
+cell types.                                                       establish CEL-Seq’s exact reproducibility and sensitivity at very
+                                                         low amounts of input RNA. Finally, we apply our protocol to study
+                                                                          sister cells from early C. elegans embryos, and demonstrate thatINTRODUCTION
+                                                       CEL-Seq’s high performance can be used to reliably distinguish
+                                                      between cell types, even in cases where only subtle biologicalFor many biological questions a single-cell-level description of
+                                                                  differences are present.gene regulation  is advantageous to  cell populations (Tang
+et al., 2011; Wang and Bodovitz, 2010). Microscopy, FACS, or
+real-time PCR-based methods can provide a single-cell aspect  RESULTS
+to experiments but are able to assay only a handful of genes at
+a time. High-throughput technologies such as microarrays and  CEL-Seq Performs Multiplexed Single-Cell
+RNA-Seq provide a full view of the expression of all genes but   Transcriptomics by Linear Amplification
+are limited by the amount of RNA needed for analysis. This can   The CEL-Seq method begins with a single-cell reverse-tran-
+be solved by adding an RNA amplification step, either by expo-   scription reaction using a primer designed with an anchored
+nential PCR-based amplification or linear in vitro transcription   polyT, a unique barcode, the 50 Illumina sequencing adaptor,
+(IVT) amplification (Eberwine et al., 1992). With PCR practically   and a T7 promoter (Figure 1A; see Experimental Procedures
+any RNA starting amount can be employed, simply by adding    for details). Next, second-strand synthesis is performed and
+additional cycles, thereby allowing analysis at the single-cell   then the cDNA samples are pooled and consequently comprise
+level. However, efforts for linear amplification of RNA from single   sufficient template material for an IVT reaction. The amplified
+cells have been challenged by IVT’s lower bound of  400 pg  RNA is then subjected to directional RNA library preparation.
+total RNA as input material for a single round of amplification.   The RNA is fragmented to a size distribution appropriate for
+Therefore, to date, IVT has not been efficiently used for amplifi-   sequencing, the Illumina 30 adaptor is added by ligation, RNA
+cation of RNA from single cells (Tang et al., 2011).                     is reverse transcribed to DNA, and the 30-most fragments that
+
+
+
+666  Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors
+
+## Page 2
+
+A
+
+
+
+
+
+                                 C                         E
+
+
+
+
+
+B
+
+                                 D
+
+
+
+
+
+Figure 1. The CEL-Seq Method
+(A) Individual cells are added to tubes, each with a uniquely bar-coded primer for reverse transcription. After second-strand synthesis, the reactions are pooled for
+IVT. The amplified RNA is then fragmented and purified before entry into a modified version of the Illumina directional RNA protocol, the molecules with both
+Illumina adaptors are selected, and the DNA library is sequenced with paired-end reads.
+(B) Nucleotide distribution in the sequenced paired-end reads. Each nucleotide position is represented by one column, with the first base on the left.
+(C) Barcode distribution of one IVT reaction after demultiplexing. The cells from three two-cell stage C. elegans embryos (denoted P1 and AB) and a single one-cell
+stage embryo (denoted P0) were amplified together in a single multiplexed IVT reaction.
+(D) Distribution of the reads mapping to the C. elegans genome in the six AB/P1 cells. Error bars indicate the SD.
+(E) Correlation between biological AB replicates.
+See also Figure S1C.
+
+
+
+contain both Illumina adaptors and a barcode are selected. The   precisely at the beginning of the first read, invariably followed
+resulting library undergoes paired-end sequencing, where the   by a polyT stretch (Figure 1B). Barcodes from all six samples
+first read recovers the barcode, whereas the second identifies   were represented,  indicating the success  of the  individual
+the mRNA transcript (Figure 1A). Thus, by multiplexing CEL-   single-cell reverse-transcription reactions (Figure 1C). We map-
+Seq takes advantage of the different input requirements of the  ped reads to the C. elegans genome and found that 91.7%
+reverse-transcription and IVT reactions to obtain sufficient RNA  stemmed from mRNA. Only 2.0% stemmed from ribosomal
+from single cells for a single round of linear amplification.      RNA (rRNA), demonstrating CEL-Seq’s specificity for polyadeny-
+  As an initial test, we applied CEL-Seq to individual cells iso-   lated transcripts. This was also supported by the extremely low-
+lated from three two-cell C. elegans embryos (Table S1). On   detected expression levels of core histone mRNAs, which are
+average, 95.5% of the filtered reads had a barcode located   thought to be highly expressed yet mostly nonpolyadenylated
+
+
+
+                                                                      Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors  667
+
+## Page 3
+
+A                    B                           C
+
+
+
+
+
+Figure 2. Benchmarking of CEL-Seq on Mouse ES and MEF Cells
+(A) CEL-Seq Pearson’s correlation coefficients among the ES and MEF cells (on log10 tpm values), computed as previously described by Islam et al. (2011), on the
+1,000 genes most highly expressed in ES cells and 1,000 genes most highly expressed in MEF cells (1,385 genes).
+(B) Mean number of genes detected above two thresholds (10 and 100 tpm) across the ES and MEF cell types using CEL-Seq and the ‘‘STRT’’ PCR-based
+method (Islam et al., 2011). Error bars indicate 95% confidence intervals.
+(C) Reproducibility according to expression level. For each gene the coefficient of variation was computed across the log10 tpm values in the ES cells for the STRT
+and CEL-Seq methods. The genes were then ranked by expression level in bins of 200 from high to low. For each bin the mean and SD of the coefficients of
+variation of the genes are shown.
+See also Figure S2 for additional analyses.
+
+
+(Figure  S1A).  Finally, RNase  treatment  of  cells  did  not   (Figure 2B). In order to ensure a fair comparison between the
+produce amplified RNA indicating the specificity of the method   two methods, we quantified reproducibility for each method
+to RNA.                                                           separately, using the same criteria as previously described by
+  CEL-Seq is highly strand specific because 97.2% of exonic   Islam et al. (2011), and found that with CEL-Seq, significantly
+reads exhibited sense orientation (Figure 1D). The reads mapped   lower noise was detected across biological replicates for both
+exclusively to the 30 end of transcripts (Figure S1B), which was    cell types tested (Figures 2C and S2C). Finally, principal compo-
+expected because CEL-Seq only retains the 30-most fragments   nent analysis based on CEL-Seq data better distinguishes
+of transcripts (Figure 1A). Some reads mapped to intergenic   between cell types than the corresponding STRT data (Fig-
+sequences, but manual inspection of the aligned reads revealed   ure S2D). Biological variation between replicates is a confound-
+that this can likely be explained by incomplete 30 UTR annota-   ing factor when trying to establish the performance of a method.
+tions (data not shown). Expression levels were then estimated  We therefore also compared the two methods based on expres-
+by counting all reads mapping to each gene, and normalized to   sion levels of exogenously introduced RNA (see below) and
+give the read count in transcripts per million (tpm; see Experi-   found that CEL-Seq provided more reproducible measurements
+mental Procedures). The expression levels of all genes (hence-   (Figure S2E).
+forth, transcriptome) across biological replicates showed an
+average correlation of R = 0.979 (Figures 1E and S1C). A nega-  CEL-Seq Is Highly Sensitive and Reproducible
+tive control—starting with the growth media lacking a cell—   In order to determine CEL-Seq’s sensitivity and reproducibility,
+resulted in very few reads (Figure S1D).                   we analyzed different amounts of purified C. elegans RNA from
+                                                      mixed  embryonic  stages,  eliminating  biological  variability
+CEL-Seq Outperforms a PCR-Based Multiplexed RNA-      present between single cells and allowing us to use different dilu-
+Seq Method                                                       tions of the same RNA. We prepared stepwise dilutions, from
+We next sought to compare the performance of our protocol to   40 pg of total RNA down to levels representative of mammalian
+the STRT method, a previously introduced PCR-based multi-   single cells ( 5 pg). In parallel we sequenced a 1 ng RNA sample
+plexed RNA-Seq method by Islam et al. (2011). We thus applied   from the same preparation for use as a reference. In half of the
+CEL-Seq to the cell types compared in this previous study by   samples, we added exogenous ‘‘carrier’’ RNA to test whether
+Islam et al. (2011) and determined the transcriptomes of nine   the overall amount of RNA in a reverse-transcription reaction
+mouse embryonic stem (ES) cells and seven mouse embryonic   affects the efficiency of the reverse-transcription step and found
+fibroblasts (MEFs) (Table S1). When comparing the distribution   that it did not: the number of reads that mapped to the C. elegans
+of expression levels of each single-cell transcriptome across  genome depended only on the amount of C. elegans RNA
+methods and  cell type, CEL-Seq shows more reproducible   present in a given sample (Figure S3A). Each sample also con-
+distributions of expression (Figure S2A). We found that CEL-   tained a set of 92 spike-in RNAs with defined concentrations,
+Seq produced higher correlations for ES cells and distinguished   spanning more than five orders of magnitude (Baker et  al.,
+between cell types more clearly, when examining the highly ex-   2005), which showed a linear response across the entire detec-
+pressed genes in either cell type (Figures 2A and S2B). Further-   tion range (R2 = 0.87 ± 0.04 for the 10 pg samples; Figures 3A
+more, CEL-Seq detected significantly more genes in the ES cells   and S3B).
+
+
+
+668  Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors
+
+## Page 4
+
+A                          B                                   Figure 3. Sensitivity and Reproducibility of
+                                                                           CEL-Seq
+                                                                                                                            (A) CEL-Seq achieves a linear response over the
+                                                                                                                    entire detection range. The plot indicates the 92
+                                                                       ERCC (Baker et al., 2005) spike-in levels from six
+                                                                                       10 pg replicate samples. Averages for each of the
+                                                                                       17 groups of spike-ins with the same nominal
+                                                                                                      concentration are shown as larger circles. Foreach
+                                                                                         sample the spike-ins were normalized such that
+                                                                                                       the average expression of the 12th spike-in group
+                                                                                                        containing 1,000 molecules was set to 1,000. The
+                                                                                                                  fraction of spike-ins in each group without de-
+                                                                                                      tected expression is shown at the bottom of the
+                                                                                                       figure. The line indicates an idealized linear rela-
+                                                                                                                  tionship. Systematic deviations from the line for
+C                                                                            some spike-ins with high concentrations are likely
+                                                                                     due to differences in G/C content.
+                                                                                                                        (B) C. elegans transcript counts per sample based
+                                                                                      on linear regression of spike-in expression levels
+                                                                                                                       (5 pg, n = 5; 10 pg, n = 6; 20 pg, n = 5; 40 pg, n = 3).
+                                                                                       The number of molecules calculated is directly
+                                                                                                           proportional to the amount of input RNA. Error bars
+                                                                                                             indicate the SD.
+                                                                                                                     (C) CEL-Seq sensitivity. For different amounts of
+                                                                                                            input RNA, the bars indicate the percentage of
+                                                                                         genes detected (at least one read) as a function
+                                                                                                                of their absolute copy number, calculated based
+                                                                                      on the 1 ng reference sample. Error bars indicate
+                                                                                                       the SDs.
+                                                                                    See Figure S3 for additional analyses.
+
+
+
+
+                                                                                            replicates of 20 pg C. elegans RNA. We
+                                                                                   simulated  different sequencing depths
+                                                                       and found  that  at one  million reads,
+                                                         91%  of  the genes  with an average
+ We next invoked the spike-ins to assess CEL-Seq’s reproduc-   expression >100 tpm had an expression level within 20% of
+ibility and sensitivity based upon absolute molecule counts.   that of the averaged expression and that sequencing deeper
+These calculations indicated that 10 pg of the C. elegans RNA   did not further improve reproducibility. A total of 250,000 reads
+contained approximately 280,000 mRNA molecules (Figure 3B).   produced results that were already sufficient for good quantifica-
+Using that number, and the relative expression level of each   tion and very similar to an order of magnitude higher sequencing
+gene in the reference sample, we calculated the expected   depth (Figure S3D).
+number of transcripts per gene in each of our dilution series
+and binned the genes in each sample according to this number.   Transcriptomic Analysis of Single Cells in the Early
+Comparison of the different dilutions showed that CEL-Seq’s   C. elegans Embryo Identifies Differential and New
+absolute sensitivity does not decrease with smaller starting   Expression
+amounts of RNA (Figures 3C and S3C). For example 50% of all  We next asked whether CEL-Seq can be used to identify biolog-
+genes with four to five copies and virtually all genes present in    ical differences between closely related cells in the C. elegans
+more than 50 copies per cell were detected, independent of   embryo.  C.  elegans  embryonic  development  begins  with
+the total amount of RNA analyzed (Figure 3C). Similarly, the   unequal cleavages producing founder cells—termed blasto-
+absolute reproducibility of the method is also not compromised  meres—some of which are depicted in Figure 4B. The AB and
+by smaller starting amounts (Figure S3C). In summary, CEL-   P1 sister blastomeres were examined 10 min after cell division.
+Seq’s performance  in estimating a  transcript’s abundance   Examining their transcriptomes, we detected 17 genes with
+depends solely on its absolute copy number in the sample.   a mean 2-fold difference showing significantly different expres-
+Thus, whereas the IVT imposes a minimal starting amount of   sion (p < 0.05, t test, FDR-corrected; Figure 4A). The most highly
+ 400 pg total RNA for sufficient yield in a single round, the   expressed of these is mex-3, whose differential expression is
+reverse-transcription reaction does not have this restriction.   supported by previous work by Draper et al. (1996). Shuffling
+CEL-Seq thus exploits this difference by pooling reverse-tran-   the groupings of the triplicates resulted in no differentially ex-
+scription reactions together to arrive at the IVT threshold.        pressed genes. Next, we sought to proceed in developmental
+ We also assessed the required sequencing depth for accurate   time, using CEL-Seq to further characterize the early embryonic
+transcriptomic data using CEL-Seq. We created 12 technical   transcriptome. We collected  cells by comparing the germ
+
+
+
+                                                                      Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors  669
+
+## Page 5
+
+      A                                    B
+
+
+
+
+
+      C                                     D
+
+
+
+
+
+Figure 4. Dissecting the Early C. elegans Embryo with CEL-Seq
+(A) Differential expression analysis in the two-cell stage blastomeres, AB and P1. A t test was made for the 137 genes for which there was expression >100 tpm
+and at least 2-fold change between the means of the triplicates. The 17 genes with p < 0.05 (FDR corrected) are shown along with the mean expression (right) and
+standardized expression of triplicates on the left (mean subtracted and SD divided).
+(B) The blastomeres examined in this study are abstracted in the cell lineage. The number of new transcripts is indicated under the vertical arrows. The two-sided
+arrows indicate the number of differentially expressed genes in the anterior versus the posterior blastomeres, respectively. A list of all differentially expressed and
+new transcripts is in Tables S2 and S3, respectively.
+(C) Gene expression levels (log10 tpm; see color scale on right) for the indicated genes; cell lineage is as in (B).
+(D) Classification of the AB and P1 blastomeres. For the indicated number of replicates used in the training data, the performance of the machine-learning
+classifier (see Experimental Procedures) in assigning blastomere identities is shown as a function of the number of genes included for prediction.
+See Figure S4 for additional analyses.
+
+
+lineage to the somatic cells: P1 was allowed to divide to its two    ingly, newly expressed EMS genes are enriched for transcription
+daughter cells, P2 and EMS, which were analyzed, and similarly,   factors (4 out of 17; p < 10  3, hypergeometric distribution): med-1
+P2 was allowed to divide to P3 and C, which were then analyzed.   and three ccch genes—ccch-2, F38C2.7, and Y116A8C.19—not
+We also collected the AB daughter cells. We assayed for differ-   previously known to be expressed this early in the embryo.
+ential distribution of transcripts in these additional cells similarly    The C and P3 cells are born at the eight-cell stage. Again, the
+to the AB/P1 analysis (Table S2), as well as asking which genes   P-lineage  cell has new expression of fewer genes than  its
+are newly expressed by comparing with the mother cell (Fig-   somatic sister. The C cell expresses more new genes than
+ure 4B; Table S3). The differential distribution between AB/P1  EMS, suggesting that additional repressors are removed with
+and EMS/P2 has also been tested by microarray and regular   the progression of development. Importantly, the C and EMS
+RNA-Seq, in both cases using pooled cells, showing good corre-   newly transcribed genes share a highly significant overlap (ten
+lation (Figure S4A). New transcription in a particular cell type was   genes; p < 10  27, hypergeometric distribution; see Figure S4B).
+scored when the transcript’s concentration exceeded 10 tpm   Out of the 35 C genes, 6 are transcription factors (three times
+and was <1 tpm in the mother cell (median across replicates).  more than expected; p < 10  3, hypergeometric distribution).
+We found that in the four-cell stage, only the EMS cell has   This suggests a somatic program evenly kick started upon diver-
+considerable expression of new transcripts. Among these are   gence from P-lineage transcriptional repression, likely by PIE-1
+med-1 and pes-10, known to be newly expressed in this cell   (Seydoux et al., 1996), and a crucial role for specific transcription
+(Maduro et al., 2001; Seydoux et al., 1996) (Figure 4C). Interest-   factors in early development. Finally, we found early expression
+
+
+
+670  Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors
+
+## Page 6
+
+of 13 genes in P3 (Table S3); of these, one—scrm-4—is a target   tiveness of CEL-Seq. Furthermore, the protocol is complete
+of deps-1, a P granule-associated protein (Spike et al., 2008),   from start to end: the amplification, library construction, and
+suggesting that CEL-Seq  is indeed detecting expression in   downstream bioinformatics analysis are seamlessly connected.
+a cell previously thought to be transcriptionally inert (Seydoux   The amplification can be done for up to 50 cells (samples) per
+et al., 1996).                                            day by a single person. The bar-coded primers are simple to
+   Finally, we were interested in whether transcriptomic profiles   create and manufacture. Because the barcode is 8 bp in length
+obtained with CEL-Seq can be used to computationally predict   and can be longer, the number of samples that may have unique
+the identity of unknown cells, as can be the case when analyzing   barcodes in a given IVT is essentially unlimited. These converge
+cells from complex tissues. We reasoned that distinguishing   to a single sample ready to interface with library preparation
+between the closely related C. elegans sister blastomeres would   enabling the library preparation for  10 of these, or  500 cells.
+pose a veritable challenge and therefore constitute a suitable   The library construction kit is a standard Illumina kit that itself
+test case. In order to achieve additional statistical power, we   provides an additional barcode for each library such that multiple
+analyzed AB and P1 blastomeres from six more embryos. We   IVTs can be analyzed together on the same sequencing lane.
+built a classifier that uses a training set of sister blastomeres to    Finally, we provide our analysis pipeline ready for integration
+choose a specified number of maximally informative genes   within the Galaxy framework, such that expression values can
+(Experimental Procedures). We then used this information to   easily be obtained within a matter of hours.
+predict the likely identities of new pairs of sister blastomeres.      In addition to working with single cells, CEL-Seq comes with
+Surprisingly, in assessing the performance of our classifier using   several other desirable properties, such as strand specificity
+cross-validation (Figures 4D and S4C), we found that for both the  (>98% of exonic reads come from the sense strand) and barcod-
+two- and the eight-cell stage sisters, predictions can be made   ing efficiency (>96% of the reads contain barcodes). After ampli-
+with an average 80%–90% success rate based on data from   fication, the CEL-Seq protocol selects for the single 30-most
+only three embryos. Using four or five replicates, the success   fragment of each transcript. In contrast to virtually  all other
+rate increased to 90%–100%. For EMS and P2 the classifier still  RNA-Seq methods,  this greatly simplifies the estimation of
+did far better than guessing, but the lack of high success rates   expression levels because no normalization by gene length is
+even with five replicates indicated that a systematic bias or   necessary. Thus, it will be of interest to invoke CEL-Seq when-
+outlier effect was present in the data. Still, the good success in   ever RNA amplification is necessary, even when not working
+highly similar blastomeres underscores the potential of our   with single cells.
+method for future transcriptomic applications.              We found that CEL-Seq outperformed STRT, a previously
+                                                              introduced  PCR-based  multiplexed   single-cell  RNA-Seq
+DISCUSSION                                         method (Islam et al., 2011), in terms of robustness, sensitivity,
+                                                     and reproducibility, and suffered from significantly less technical
+Single-cell transcriptomics is poised to revolutionize biological   noise. We note that the two methods examine different ends of
+research and medical practice (Tang et al., 2011; Wang and Bod-   the mRNA transcript: 50 for STRT, but 30 for CEL-Seq. In addition
+ovitz, 2010), allowing for unbiased and comprehensive cell char-   there may have been unavoidable differences in culturing condi-
+acterization. It is acknowledged, that whenever possible, linear   tions of the  cell types analyzed. However, several lines of
+amplification by IVT is preferable to exponential amplification   evidence indicate that it is unlikely that these aspects account
+by PCR (Tang et al., 2011). Here, we describe a protocol that    for the observed performance differences: (1) expression levels
+combines the power of linear amplification by IVT, with a pooling   were not compared directly across the two methods, (2) signifi-
+procedure that allows the efficient analysis of many samples in   cant differences were found for both cell types, and (3) methods
+parallel. We presented evidence that CEL-Seq is a sensitive,   were also compared based on spike-ins, which are not affected
+accurate, and reproducible single-cell transcriptomics method.   by these factors.
+We tested CEL-Seq on mammalian cells and nematode embry-    The limitations of CEL-Seq fall into the categories of specificity
+onic blastomeres, made extensive use of a suite of spike-ins, and   to mRNA, 30 bias, and sensitivity to small copy numbers. CEL-
+established the exact  sensitivity and  reproducibility  of the  Seq does not detect miRNAs and other nonpolyadenylated tran-
+method using extremely low amounts of purified RNA. Here,   scripts. This can be seen as an advantage because the bar-
+we review the advantages and limitations of the method and  coded transcripts are largely depleted of rRNA (<2%), which
+finally consider CEL-Seq’s possible applications.                 increases the efficiency of the sequenced reads to measure
+  CEL-Seq’s key advantage over other protocols (Islam et al.,  mRNA levels. Due to its strong 30 bias, the method is severely
+2011; Tang et al., 2009) arises from its ability to harness the   limited  in  its  ability to distinguish  alternative splice forms.
+power of IVT, providing both multiplexing and reproducibility.   Another aspect of the 30 localization of the reads is that in species
+By pooling many samples to a single IVT, a single round of ampli-   with genomes that are not well annotated, the reads will map to
+fication is sufficient, and CEL-Seq provides significantly reduced   unannotated 30 UTRs. This could be remedied to some extent by
+hands-on time both for the amplification and downstream pro-    artificially extending transcript annotations beyond the anno-
+cessing, allowing for the preparation of dozens of samples for   tated 30 end. Finally, a crucial issue with any single-cell gene
+sequencing within 2–3 days. CEL-Seq makes use of commer-   expression method is the sensitivity to the detection of lowly ex-
+cially available kits for the amplification and sequencing library   pressed genes. We have calculated that if the transcript is at five
+preparation; only the bar-coded primers and a few enzymes   copies, there is a 50% chance of its identification by CEL-Seq.
+need to be obtained separately. This makes for the cost-effec-   Relative to RNA-Seq of pooled samples, this may seem less
+
+
+
+                                                                      Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors  671
+
+## Page 7
+
+sensitive; however,  pooling  effectively  increases  the copy   by the total number of counted reads and multiplying with 106. Because CEL-
+number that we have shown to be the important parameter for   Seq retains one fragment per transcript, this procedure yields the estimated
+detection. Nevertheless, CEL-Seq on single cells will capture   gene expression levels in tpm. Absolute copy numbers were obtained by first
+                                                                          performing least-squares linear regression on the spike-in values. The result-
+variation in the expression levels among cells.
+                                                                                ing factor was used to convert the tpm values to mRNA copy numbers (scripts
+                                                                                   available at yanailab.technion.ac.il).
+EXPERIMENTAL PROCEDURES
+
+                                                                     Classification of Blastomere Identities
+Single-Cell Isolation                                        A machine-learning classifier was devised to predict the identities of pairs of
+C. elegans blastomeres were isolated as previously described by Edgar (1995).    sister blastomeres, and its performance was assessed using cross-validation.
+Mammalian cells were obtained by trypsin treatment of adherent cells; see also    Briefly, for a given number of samples to be used as training data, and a given
+the Extended Experimental Procedures. Individual cells (or media without a cell   number of genes to be used in the prediction (G), the classifier first ranks genes
+for negative control) were transferred with a micropipette into a 0.5 ml drop of   according to the significance of their differential expression in the training data
+egg salts or PBS for C. elegans blastomeres or mammalian cells, respectively,    (using a paired t test). The G-most different genes are selected, and their mean
+placed on the cap of a 0.5 ml LoBind Eppendorf tube, excess liquid was aspi-    differences between the classes are normalized by their SD, to yield a refer-
+rated off, and frozen in liquid nitrogen. Samples were stored at  80 C.        ence vector. For a new set of sister blastomeres, the score for the two possible
+                                                                                classifications is calculated as the Euclidean distance between the differences
+CEL-Seq Primer Design                                                 normalized by the SD from the training data, and the reference vector. The pre-
+The reverse-transcription primer was designed with an anchored polyT,    dicted classification is chosen according to the smaller distance. The number
+a unique barcode, the 50 Illumina adaptor, and a T7 promoter. The T7 promoter    of possible combinations of data sets to choose for the training step is N
+sequence was as previously described by Baugh et al. (2001). The Illumina 50   choose k, where k is the number of embryos to be used, and N is the total
+adaptor sequence was as used in the Illumina small RNA kit. The barcodes   number of embryos analyzed. For each k, all such possibilities are tested,
+were of length eight and designed in groups of four, such that the first five   and the average success rate is reported.
+nucleotides will have equal representation of all four nucleotides to allow for
+template generation and crosstalk corrections that are based on the first
+                                                     ACCESSION NUMBERS
+four nucleotides read in the Illumina platform. The barcodes were designed
+such that each pair is different by at least two nucleotides, so that a single
+                                                                 The NCBI SRA accession number for the sequence data reported in this paper
+sequencing error will not produce the wrong barcode. All used primers are
+                                                                                                    is SRP014672.
+described in the detailed Extended Experimental Procedures.
+
+                                                   SUPPLEMENTAL INFORMATIONLinear mRNA Amplification
+Ambion’s MessageAmp  II aRNA Kit (AM1751) was used with the following
+                                                                      Supplemental Information includes Extended Experimental Procedures, fourmodifications. The polyT primer was replaced with the CEL-Seq primer. The
+                                                                            figures, three tables, and detailed protocol of the CEL-Seq method and canreverse-transcription reaction was performed at one-tenth volume, with 5 ng
+                                                                be found with this article online at http://dx.doi.org/10.1016/j.celrep.2012.of primer per reaction. A total of 0.2 ml of the primer mixed with 1 ml of water
+or 1 ml of a 1:500,000 dilution of the ERCC spike-in kit (a total of 1.2 ml) was    08.003.
+added directly to the lid of the Eppendorf tube where the cell was frozen,
+and incubated at 70 C for 10 min (with the lid of the thermal cycler heated to   LICENSING INFORMATION
+70 C). The sample was spun to the bottom of the tube midincubation. After
+the second-strand synthesis, samples were pooled and cleaned on a single    This is an open-access article distributed under the terms of the Creative
+column before proceeding to the IVT reaction at two-fifths volume for 13 hr.   Commons Attribution-Noncommercial-No Derivative Works 3.0 Unported
+RNA was fragmented (one-fifth volume of 200 mM Tris-acetate [pH 8.1],    License (CC-BY-NC-ND; http://creativecommons.org/licenses/by-nc-nd/3.0/
+500 mM KOAc, 150 mM MgOAc added) for 3 min at 94 C, and the reaction    legalcode).
+was stopped by placing on ice and the addition of one-tenth volume of
+0.5 M EDTA, followed by RNA cleanup. The RNA quality and yield were as-  ACKNOWLEDGMENTS
+sayed using a Bioanalyzer (Agilent).
+                                                                              This work was supported by the Israel Science Foundation, an FP7 IRG grant
+Library Construction and Sequencing                               and the Lorry  I. Lokey Interdisciplinary Center for Life Sciences and Engi-
+Illumina’s directional RNA sequencing protocol was used with the following    neering at the Technion. We acknowledge the Gepstein and Aberdam labora-
+modifications. A total of 5 ng of RNA was used as input. The mRNA pull-    tories at the Technion for assistance with the mammalian cells. We also thank
+down and fragmentation steps were skipped because amplified RNA repre-    Daniel Glikman for a critical reading and advice. T.H., N.S., and I.Y. conceived
+sents only mRNA sequences and was already fragmented. Only the 30 Illumina    the method. N.S. led the development of the method. T.H. isolated the blasto-
+adaptor was ligated—diluted 1:5 prior to ligation to obtain the appropriate   meres and mouse cells, performed the validation with other methods, and
+molar ratio with the reduced amount of RNA. A total of 12 cycles of PCR   managed the DNA sequencing. F.W. developed the pipeline to derive the
+was performed with an elongation time of 30 s. Libraries were sequenced on   gene expression levels and performed the quality controls. T.H., F.W., and
+the  Illumina  HiSeq2000  according  to  standard  protocols.  Paired-end     I.Y. analyzed the data. All authors contributed to the experimental designs
+sequencing was performed, reading at least 15 bases for read 1, and 50 bases   and the writing of the manuscript.
+for read 2, and the Illumina barcode when needed.
+                                                                         Received: June 12, 2012
+Expression Analysis Pipeline                                             Revised: July 18, 2012
+Transcript abundances were obtained from the sequencing data using custom   Accepted: August 3, 2012
+scripts organized into a multistep, paralleled computational pipeline within the   Published online: August 30, 2012
+Galaxy framework (Giardine et al., 2005). Briefly, after trimming and filtering,
+the paired-end reads were demultiplexed based on the first eight bases of  REFERENCES
+the first read. For each sample, reads were mapped to the C. elegans reference
+genome (WS230; http://www.wormbase.org), counted using htseq-count    Baker, S.C., Bauer, S.R., Beyer, R.P., Brenton, J.D., Bromley, B., Burrill, J.,
+(http://www-huber.embl.de/users/anders/HTSeq), and normalized by dividing   Causton, H., Conley, M.P., Elespuru, R., Fero, M., et al; External RNA Controls
+
+
+
+672  Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors
+
+## Page 8
+
+Consortium. (2005). The External RNA Controls Consortium: a progress report.   Maduro, M.F., Meneghini, M.D., Bowerman, B., Broitman-Maduro, G., and
+Nat. Methods 2, 731–734.                                             Rothman, J.H. (2001). Restriction of mesendoderm to a single blastomere
+                                                                  by the combined action of SKN-1 and a GSK-3beta homolog is mediated by
+Baugh, L.R., Hill, A.A., Brown, E.L., and Hunter, C.P. (2001). Quantitative anal-
+                                                         MED-1 and -2 in C. elegans. Mol. Cell 7, 475–485.
+ysis of mRNA amplification by in vitro transcription. Nucleic Acids Res. 29, E29.
+                                                                     Seydoux, G., Mello, C.C., Pettitt, J., Wood, W.B., Priess, J.R., and Fire, A.
+Draper, B.W., Mello, C.C., Bowerman, B., Hardin, J., and Priess, J.R. (1996).
+                                                                                    (1996). Repression of gene expression in the embryonic germ lineage of
+MEX-3 is a KH domain protein that regulates blastomere identity in early
+                                                                        C. elegans. Nature 382, 713–716.
+C. elegans embryos. Cell 87, 205–216.
+                                                                             Spike, C.A., Bader, J., Reinke, V., and Strome, S. (2008). DEPS-1 promotes
+Eberwine, J., Yeh, H., Miyashiro, K., Cao, Y., Nair, S., Finnell, R., Zettel, M., and    P-granule assembly and RNA interference in C. elegans germ cells. Develop-
+Coleman, P. (1992). Analysis of gene expression in single live neurons. Proc.   ment 135, 983–993.
+Natl. Acad. Sci. USA 89, 3010–3014.
+                                                                       Tang, F., Barbacioru, C., Wang, Y., Nordman, E., Lee, C., Xu, N., Wang, X.,
+Edgar, L.G. (1995). Blastomere culture and analysis. Methods Cell Biol. 48,   Bodeau, J., Tuch, B.B., Siddiqui, A., et al. (2009). mRNA-Seq whole-transcrip-
+303–321.                                                        tome analysis of a single cell. Nat. Methods 6, 377–382.
+
+Giardine, B., Riemer, C., Hardison, R.C., Burhans, R., Elnitski, L., Shah, P.,   Tang, F., Lao, K., and Surani, M.A. (2011). Development and applications of
+Zhang, Y., Blankenberg, D., Albert, I., Taylor, J., et al. (2005). Galaxy: a platform    single-cell transcriptome analysis. Nat. Methods 8 (4, Suppl), S6–S11.
+for interactive large-scale genome analysis. Genome Res. 15, 1451–1455.     Wang, D., and Bodovitz, S. (2010). Single cell analysis: the new frontier in
+Islam, S., Kja¨llquist, U., Moliner, A., Zajac, P., Fan, J.B., Lo¨nnerberg, P., and    ‘omics’. Trends Biotechnol. 28, 281–290.
+Linnarsson, S. (2011). Characterization of the single-cell transcriptional land-   Wang, Z., Gerstein, M., and Snyder, M. (2009). RNA-Seq: a revolutionary tool
+scape by highly multiplex RNA-seq. Genome Res. 21, 1160–1167.                for transcriptomics. Nat. Rev. Genet. 10, 57–63.
+
+
+
+
+
+                                                                      Cell Reports 2, 666–673, September 27, 2012 ª2012 The Authors  673

cel_seq/CEL-Seq_protocol.docling_text.txt

@@ -0,0 +1,628 @@
+# docling text extraction
+source_pdf: CEL-Seq_protocol.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/cel_seq/CEL-Seq_protocol.pdf
+extraction: Docling: PdfPipelineOptions(do_ocr=False); assembled markdown plus Docling native PDF layout prediction text cells sorted by page/y/x
+
+## Docling assembled markdown
+
+## CEL-Seq Protocol
+
+## By Tamar Hashimshony, Technion July 16 th , 2012
+
+## Reagents:
+
+LoBind tubes - 0.5 ml - Eppendorf 022431005 Ultra pure RNase free water Ethanol Bioanalyzer kits - Agilent RNA pico kit (5067-1513), high sensitivity DNA kit (5067-4626) Qubit reagents: dsDNA HS Assay - invitrogen Q32851 or Q32854 For RNA amplification: ERCC RNA spike-in mix - Ambion 4456740 MessageAmpII kit - Ambion AM1751 Optional: extra columns for cDNA/aRNA purification - Ambion10066G Fragmentation buffer: 200mM Tris-acetate, pH 8.1, 500 mM KOAc, 150 mM MgOAc Fragmentation stop buffer: 0.5 M EDTA For Library preparation: Antarctic phosphatase - NEB M0289 RNase OUT - invitrogen 100000840 PNK - NEB M0201 Superscript II - invitrogen 18064-014 RNeasy MinElute kit - Qiagen 74204 T4 RNA ligase 2, truncated - NEB M0242 AMPure XP beads - Beckman Coulter A63880 TruSeq small RNA sample prep kit - Illumina RS-200-0012 (or -0024, -0036, -0048) Optional (to supplement Illumina's Small RNA kit): ATP - NEB P0756 Phusion® High-Fidelity PCR Master Mix with HF Buffer - NEB M0531 RNA RT primer, RNA PCR primers (sequences available from Illumina)
+
+## Equipment:
+
+Thermocycler with lid with adjustable temperature (one that can also fit 0.5 ml PCR tubes is convenient) Speed vac Oven Heat block Magnetic stand (for 0.5 ml tubes) Qubit® Fluorometer - invitrogen Bioanalyzer - Agilent
+
+## Primers:
+
+CEL-Seq primer design: The RT primer was designed with an anchored polyT, a unique barcode, the 5' Illumina adapter (as used in the Illumina small RNA kit) and a T7 promoter. The barcodes were of length eight and designed in groups of four, such that the first five nucleotides will have equal representation first four nucleotides read in the Illumina platform. The barcodes were designed such that each pair is different by at least two nucleotides, so that a single sequencing error will not produce the wrong
+
+of all four nucleotides to allow for template generation and crosstalk corrections which are based on the barcode. Primers are desalted, stock solution 1 μg/μl, working concentration 25ng/μl. #1:  CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCACGCTTTTTTTTTTTTTTTTTTTTTTTTV #2: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV #3: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACCGCTTTTTTTTTTTTTTTTTTTTTTTTV #4: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGCGCTTTTTTTTTTTTTTTTTTTTTTTTV #5: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAATCTTTTTTTTTTTTTTTTTTTTTTTTV #6: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTATCTTTTTTTTTTTTTTTTTTTTTTTTV #7: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACATCTTTTTTTTTTTTTTTTTTTTTTTTV #8: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGATCTTTTTTTTTTTTTTTTTTTTTTTTV #9: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCATCCTTTTTTTTTTTTTTTTTTTTTTTTV #10: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTTCCTTTTTTTTTTTTTTTTTTTTTTTTV #11: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACTCCTTTTTTTTTTTTTTTTTTTTTTTTV #12: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGTCCTTTTTTTTTTTTTTTTTTTTTTTTV #13: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAGAATTTTTTTTTTTTTTTTTTTTTTTTV #14: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTGAATTTTTTTTTTTTTTTTTTTTTTTTV #15: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACGAATTTTTTTTTTTTTTTTTTTTTTTTV #16: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGGAATTTTTTTTTTTTTTTTTTTTTTTTV #17: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACACGCTTTTTTTTTTTTTTTTTTTTTTTTV #18: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV #19: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACCGCTTTTTTTTTTTTTTTTTTTTTTTTV #20: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV #21: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACATCATTTTTTTTTTTTTTTTTTTTTTTTV #22: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTTCATTTTTTTTTTTTTTTTTTTTTTTTV #23: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACTCATTTTTTTTTTTTTTTTTTTTTTTTV #24: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGTCATTTTTTTTTTTTTTTTTTTTTTTTV #25: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACAGAGTTTTTTTTTTTTTTTTTTTTTTTTV #26: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV #27:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACGAGTTTTTTTTTTTTTTTTTTTTTTTTV #28: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV #29: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACCGCTTTTTTTTTTTTTTTTTTTTTTTTV #30: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV #31: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCACGCTTTTTTTTTTTTTTTTTTTTTTTTV #32: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV #33: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACTCATTTTTTTTTTTTTTTTTTTTTTTTV #34: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGTCATTTTTTTTTTTTTTTTTTTTTTTTV #35: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCATCATTTTTTTTTTTTTTTTTTTTTTTTV #36: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTTCATTTTTTTTTTTTTTTTTTTTTTTTV #37: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACGAGTTTTTTTTTTTTTTTTTTTTTTTTV #38: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV #39:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCAGAGTTTTTTTTTTTTTTTTTTTTTTTTV #40: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+## Single cell isolation :
+
+Individual cells (so far we've worked with C. elegans blastomeres or trypsinised tissue culture cells) are transferred with a micro-pipette into a 0.5µl drop of appropriate buffer (egg salts or PBS) placed on the cap of a 0.5 ml LoBind Eppendorf tube. Location of cell should be marked. Excess liquid is aspirated off, and tube is frozen in liquid nitrogen. Samples are stored at -80°C.
+
+## RNA Amplification:
+
+## Prepare primer mix (for each different primer used):
+
+Primer (25ng/μl)
+
+1μl
+
+ERCC Spike-in          Xμl
+
+water
+
+Yμl
+
+6μl
+
+Spike-in dilution should be appropriate for sample size - see protocol of ERCC RNA spike in mix. For single cells we add 1ul of spike-in at 1:500,000 dilution.
+
+## Breaking cell open and annealing with primer:
+
+-  Add 1.2μl primer mix to marked location of single cell on cap of tube (Keep cell frozen until adding the primer mix, handle up to 12 cells in parallel).
+-  Incubate 5 min. at 70 o C (with lid of thermal cycler set to 70 o C).
+-  Brief spin down.
+-  Incubate for an additional 5 min. at 70 o C.
+-  Move immediately to ice.
+-  Spin at maximal speed for a few seconds to collect as many droplets as possible before next step, and then return to ice.
+
+## RT reaction (Ambion kit)
+
+-  Add 0.8μl of the following mix to each reaction:
+-  Incubate 2hr at 42 o C ( in hybridization oven)
+
+First Strand buffer
+
+0.2μl
+
+dNTP
+
+0.4μl
+
+RNase Inhibitor
+
+0.1μl
+
+ArrayScript
+
+0.1μl
+
+## Second strand reaction (Ambion kit):
+
+-  Move previous step to ice so it cools below 16 o C.
+-  Add 8uL of the following mix to each reaction tube:
+
+| DDW                | 6.3μl   |
+|--------------------|---------|
+| Second strand buf. | 1μl     |
+| dNTP               | 0.4μl   |
+| DNA Pol            | 0.2μl   |
+| RNaseH             | 0.1μl   |
+
+Flick and spin samples (at maximal speed for a few seconds).
+
+-  Incubate at 16 o C for 2hr (in thermal cycler with unheated or open lid).
+
+## cDNA cleanup and speedvac:
+
+-  Pool all cells that are to go to same IVT.  Should have ~10μl from each cell.
+-  Adjust volume to 100μl with nuclease free water.  If more than 10 cells in a pool, just add all together.
+-  Add 250μl cDNA binding buffer to each 100ul sample (if total sample volume exceeds 100uL, adjust cDNA binding buffer volume according to sample volume). Load onto Ambion cDNA cleanup column. Volume of up to 24 pooled samples can be loaded. If more than 24 samples are pooled, after spin load remaining volume and spin again.
+-  Spin 1 min at 10,000g to bind cDNA, discard flow through (repeat if more than 24 samples are pooled).
+-  Add 500μl wash buffer, spin as above, discard flow through.
+-  Spin for an additional minute to dry.
+-  Transfer column to clean round bottom 2 ml tube.
+-  Elute by adding 9μl warm water (55 o C), incubating for 2 minutes at room temperature and spinning 1.5 min at 10,000g.
+-  Repeat elution.
+-  Adjust volume to 6.4μl by drying in a speedvac (~8 min).
+
+Stopping point: Samples can be kept at -20 o C
+
+## IVT (Ambion kit):
+
+-  Prepare the following mix and add 9.6μl per tube.
+
+A 1.6μl G 1.6μl C 1.6μl U 1.6μl 10xT7 buffer 1.6μl T7 enzyme 1.6μl
+
+-  Incubate in a thermal cycler at 37 o C for 13 hrs, with lid at 70 o C.  Set cycler to go to 4 o C at end of incubation. aRNA (amplified RNA) is stable for at least several hours.
+
+## RNA fragmentation and cleanup:
+
+-  Mix the following on ice:
+
+aRNA 16μl
+
+Fragmentation buffer 4μl
+
+-  Incubate for 3 min. at 94 o C.
+-  Immediately move to ice and add 2μl fragmentation stop buffer.
+-  Adjust volume to 30ul by adding 8μl water.
+-  Add 105μl aRNA binding buffer, followed by 75μl EtOH, immediately mix by pipetting 3-4 times and load onto spin-column. Bind sample by immediately spinning for 1 min at 10,000g, discard flow through. (If cleaning more than one reaction, this step should be done for each sample separately)
+-  Add 500μl wash buffer (samples can wait at this step until binding of all samples is complete).
+-  Spin as above, discard flow through.
+-  Spin for an additional minute to dry.
+-  Transfer column to clean tube.
+-  Elute by adding 10μl warm water (55 o C), incubating for 2 minutes at room temperature and spinning 1.5 min at 10,000 g.
+-  Repeat elution.
+
+Stopping point: Samples can be kept at -80
+
+## Check aRNA amount and quality:
+
+-  Load 1μl onto Bioanalyzer RNA pico chip after heating an aliquot of the sample to 70 o for 2 min.
+-  When starting the IVT with ~0.5ng total RNA, the expected yield is 500-1000 pg/μl. Size distribution should peak at ~500 bp (See Bioanalyzer plot for example).
+
+<!-- image -->
+
+o C
+
+## Library preparation:
+
+Protocol designed for 5-10ng amplified RNA, although as little as 1-2ng can be used, but then additional PCR cycles are required. Sample volume should be adjusted to 16μl, either by adding water or drying down in a speedvac, depending on RNA concentration. IVTs can be pooled at this point if there is no overlap in barcodes used.
+
+## Phosphatase treatment:
+
+To 16μl of fragmented aRNA in a 0.7ml PCR tube add 4μl of the following mix:
+
+-  10X phospatase buffer 2μl
+
+-  Antarctic phosphatase 1μl
+
+-  RNaseOUT 1μl
+
+Incubate in a thermal cycler with the following protocol:
+
+-  37°C for 30 minutes
+-  65°C for 5 minutes
+-  4°C indefinite hold
+
+## PNK treatment:
+
+To the 0.7 ml PCR tube from the previous step add 30μl of the following mix:
+
+-  nuclease-free H2O 17μl
+
+-  10X phosphatase buffer 5μl
+
+-  ATP (10mM, from Illumina kit) 5μl
+
+-  RNaseOUT 1μl
+
+-  PNK 2μl
+
+Incubate in a thermal cycler at 37°C for 60 minutes then 4°C hold.
+
+## Column cleanup of phosphatase and PNK treated aRNA (RNeasy kit):
+
+-  Adjust the volume to 100μl (add 50μL) using nuclease free water.
+-  Add 350μl of RLT buffer and mix well.
+-  Add 250μl EtOH, mix well by pipetting, and transfer sample to an RNeasy spin column.
+-  Spin 15 sec. at 8,000 g.
+-  Transfer column to new collection tube, and add 500μl Buffer RPE.
+-  Spin 15 sec. at 8,000 g.
+-  Discard flow-through, and add 500 ml 80% EtOH.
+-  Spin 2 min. at 8,000 g.
+-  Transfer column to new collection tube, open lid of column, and spin 5 min at full speed.
+-  Transfer column to new collection tube, and elute with 14μl nuclease free H2O, spinning at full speed for 1 minute.
+-  Dry down the sample using a speedvac to 5μl. (approx. 7 minutes)
+
+## Ligate 3' adapter:
+
+Dilute 3' adapter (RA3, from Illumina kit) 5 fold.
+
+To 5μl phosphatase and PNK treated RNA add 1μl of the diluted 3' adaptor.
+
+Incubate at 70°C for 2 minutes and then immediately place the tube on ice to prevent secondary structure formation.
+
+## Add 4μl of the following mix:
+
+-  5X HM Ligation Buffer (HML, Illumina kit)
+
+2μL
+
+- 
+
+- RNase Inhibitor (Illumina kit) 1μL
+
+- 
+
+- T4 RNA Ligase 2, truncated 1μL
+
+Incubate the tube on the pre-heated thermal cycler at 28°C for 1 hour (with unheated or open lid).
+
+With the reaction tube remaining on the thermal cycler, add 1μl Stop Solution (STP, Illumina kit) and gently pipette the entire volume up and down 6-8 times to mix thoroughly. Continue to incubate the reaction tube on the thermal cycler at 28°C for 15 minutes, and then place the tube on ice.
+
+Add 3μL nuclease free water.
+
+## Reverse transcription reaction:
+
+Dilute dNTPs (from Illumina kit) two fold with nuclease free water (prepare at least 1μl per sample)
+
+Combine the following in a PCR tube (the remaining 3' adapter-ligated RNA may be stored at -80°C):
+
+-  Adapter-ligated RNA 6μL
+
+-  RNA RT Primer (RTP, from Illumina kit)   1μL
+
+Incubate the tube at 70°C for 2 minutes and then immediately place the tube on ice.
+
+Add 5.5μl of the following mix:
+
+-  5X First Strand Buffer
+
+2μL
+
+-  12.5 mM dNTP mix (diluted dNTP)
+
+0.5μL
+
+-  100 mM DTT 1μL
+
+- 
+
+- RNase Inhibitor (Illumina kit) 1μL
+
+-  SuperScript II Reverse Transcriptase
+
+1μL
+
+Incubate the tube in the pre-heated thermal cycler at 50°C for 1 hour and then place the tube on ice.
+
+## PCR amplification:
+
+To each reverse transcription reaction add 35.5μl of the following mix:
+
+-  Ultra Pure Water
+
+8.5μL
+
+-  PCR mix (PML, from Illumina kit))
+
+25μL
+
+-  RNA PCR Primer (RP1, from Illumina kit)
+
+2μL
+
+To each reaction add 2μl of a uniquely indexed RNA PCR Primer (RPIX, from Illumina kit)
+
+Amplify the tube in the thermal cycler using the following PCR cycling conditions:
+
+-  30 seconds at 98°C
+-  12 cycles of:
+-  10 seconds at 98°C
+-  30 seconds at 60°C
+-  30 seconds at 72°C
+-  10 minutes at 72°C
+-  Hold at 4°C
+
+Can go up to 15 cycles if necessary, or down to 11 if starting with the full 10ng.
+
+Stopping point: samples can be kept at -20 o C.
+
+## Bead Cleanup of PCR products - Repeat 1:
+
+-  Prewarm beads to room temperature.
+-  Vortex AMPure XP Beads until well dispersed, then add 50μl to the 50μl PCR reaction. Mix entire volume up ten times to mix thoroughly.
+-  Incubate at room temperature for 15 min.
+-  Place on magnetic stand for at least 5 min, until liquid appears clear.
+-  Remove and discard 95μl of the supernatant.
+-  Add 200μl freshly prepared 80% EtOH.
+-  Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+-  Add 200μl freshly prepared 80% EtOH
+-  Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+-  Air dry beads for 15 min, or until completely dry.
+-  Resuspend with 32.5μl Resuspension Buffer (from Illumina kit). Pipette entire volume up and down ten times to mix thoroughly.
+-  Incubate at room temperature for 2 min.
+-  Place on magnetic stand for 5 min, until liquid appears clear.
+-  Transfer 30μl of supernatant to new tube.
+
+## Bead Cleanup of PCR products - Repeat 2:
+
+Repeat as above, but adding 39μl beads and eluting in 12.5μl resuspension buffer at the end, transferring 10μl to a new tube.
+
+## Check library amount and quality:
+
+Check concentration of DNA by Qubit, 1μl should be enough to measure using the high sensitivity reagent; expected concentration is at least ~1ng/μl.
+
+Run 1μl of each sample on Bioanalyzer using a high sensitivity DNA chip to see size distribution. Expected peak at 300-400bp (See Bioanalyzer plot for example).
+
+<!-- image -->
+
+Concentration to be loaded for sequencing should be calibrated by the sequencing facility. For us, 5pM on Hi-Seq v.1 reagents gave good cluster density.
+
+## Docling layout prediction text cells
+
+### Page 1
+
+CEL-Seq Protocol
+By Tamar Hashimshony, Technion
+July 16 th , 2012
+Reagents:
+LoBind tubes - 0.5 ml - Eppendorf 022431005
+Ultra pure RNase free water
+Ethanol
+Bioanalyzer kits - Agilent RNA pico kit (5067-1513), high sensitivity DNA kit (5067-4626)
+Qubit reagents: dsDNA HS Assay - invitrogen Q32851 or Q32854
+For RNA amplification:
+ERCC RNA spike-in mix - Ambion 4456740
+MessageAmpII kit - Ambion AM1751
+Optional: extra columns for cDNA/aRNA purification - Ambion10066G
+Fragmentation buffer: 200mM Tris-acetate, pH 8.1, 500 mM KOAc, 150 mM MgOAc
+Fragmentation stop buffer: 0.5 M EDTA
+For Library preparation:
+Antarctic phosphatase - NEB M0289
+RNase OUT - invitrogen 100000840
+PNK - NEB M0201
+Superscript II - invitrogen 18064-014
+RNeasy MinElute kit - Qiagen 74204
+T4 RNA ligase 2, truncated - NEB M0242
+AMPure XP beads - Beckman Coulter A63880
+TruSeq small RNA sample prep kit - Illumina RS-200-0012 (or -0024, -0036, -0048)
+Optional (to supplement Illumina's Small RNA kit):
+ATP - NEB P0756
+Phusion® High-Fidelity PCR Master Mix with HF Buffer - NEB M0531
+RNA RT primer, RNA PCR primers (sequences available from Illumina)
+Equipment:
+Thermocycler with lid with adjustable temperature (one that can also fit 0.5 ml PCR tubes is convenient)
+Speed vac
+Oven
+Heat block
+Magnetic stand (for 0.5 ml tubes)
+Qubit® Fluorometer - invitrogen
+Bioanalyzer - Agilent
+
+### Page 2
+
+Primers:
+CEL-Seq primer design: The RT primer was designed with an anchored polyT, a unique barcode, the 5'
+Illumina adapter (as used in the Illumina small RNA kit) and a T7 promoter. The barcodes were of length
+eight and designed in groups of four, such that the first five nucleotides will have equal representation
+of all four nucleotides to allow for template generation and crosstalk corrections which are based on the
+first four nucleotides read in the Illumina platform. The barcodes were designed such that each pair is
+different by at least two nucleotides, so that a single sequencing error will not produce the wrong
+barcode. Primers are desalted, stock solution 1 μg/μl, working concentration 25ng/μl.
+#1:  CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#2: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#3: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#4: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#5: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#6: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#7: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#8: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#9: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCATCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#10: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#11: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#12: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#13: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#14: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#15: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#16: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#17: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#18: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#19: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#20: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#21: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#22: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#23: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#24: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#25: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#26: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#27:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#28: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#29: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#30: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#31: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#32: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#33: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#34: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#35: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#36: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#37: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#38: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#39:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#40: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+### Page 3
+
+Single cell isolation :
+Individual cells (so far we've worked with C. elegans blastomeres or trypsinised tissue culture cells) are
+transferred with a micro-pipette into a 0.5µl drop of appropriate buffer (egg salts or PBS) placed on the
+cap of a 0.5 ml LoBind Eppendorf tube. Location of cell should be marked. Excess liquid is aspirated off,
+and tube is frozen in liquid nitrogen. Samples are stored at -80°C.
+RNA Amplification:
+Prepare primer mix (for each different primer used):
+Primer (25ng/μl) 1μl
+ERCC Spike-in          Xμl
+water Yμl
+6μl
+Spike-in dilution should be appropriate for sample size - see protocol of ERCC RNA spike in mix. For
+single cells we add 1ul of spike-in at 1:500,000 dilution.
+Breaking cell open and annealing with primer:
+ Add 1.2μl primer mix to marked location of single cell on cap of tube (Keep cell frozen until
+adding the primer mix, handle up to 12 cells in parallel).
+ Incubate 5 min. at 70 o C (with lid of thermal cycler set to 70 o C).
+ Brief spin down.
+ Incubate for an additional 5 min. at 70 o C.
+ Move immediately to ice.
+ Spin at maximal speed for a few seconds to collect as many droplets as possible before next
+step, and then return to ice.
+RT reaction (Ambion kit)
+ Add 0.8μl of the following mix to each reaction:
+First Strand buffer 0.2μl
+dNTP 0.4μl
+RNase Inhibitor 0.1μl
+ArrayScript 0.1μl
+ Incubate 2hr at 42 o C ( in hybridization oven)
+Second strand reaction (Ambion kit):
+ Move previous step to ice so it cools below 16 o C.
+ Add 8uL of the following mix to each reaction tube:
+
+### Page 4
+
+DDW 6.3μl
+Second strand buf. 1μl
+dNTP 0.4μl
+DNA Pol 0.2μl
+RNaseH 0.1μl
+Flick and spin samples (at maximal speed for a few seconds).
+ Incubate at 16 o C for 2hr (in thermal cycler with unheated or open lid).
+cDNA cleanup and speedvac:
+ Pool all cells that are to go to same IVT.  Should have ~10μl from each cell.
+ Adjust volume to 100μl with nuclease free water.  If more than 10 cells in a pool, just add all
+together.
+ Add 250μl cDNA binding buffer to each 100ul sample (if total sample volume exceeds 100uL,
+adjust cDNA binding buffer volume according to sample volume). Load onto Ambion cDNA
+cleanup column. Volume of up to 24 pooled samples can be loaded. If more than 24 samples are
+pooled, after spin load remaining volume and spin again.
+ Spin 1 min at 10,000g to bind cDNA, discard flow through (repeat if more than 24 samples are
+pooled).
+ Add 500μl wash buffer, spin as above, discard flow through.
+ Spin for an additional minute to dry.
+ Transfer column to clean round bottom 2 ml tube.
+ Elute by adding 9μl warm water (55 o C), incubating for 2 minutes at room temperature and
+spinning 1.5 min at 10,000g.
+ Repeat elution.
+ Adjust volume to 6.4μl by drying in a speedvac (~8 min).
+Stopping point: Samples can be kept at -20 o C
+IVT (Ambion kit):
+ Prepare the following mix and add 9.6μl per tube.
+A 1.6μl
+G 1.6μl
+C 1.6μl
+U 1.6μl
+10xT7 buffer 1.6μl
+T7 enzyme 1.6μl
+ Incubate in a thermal cycler at 37 o C for 13 hrs, with lid at 70 o C.  Set cycler to go to 4 o C at end of
+incubation. aRNA (amplified RNA) is stable for at least several hours.
+
+### Page 5
+
+RNA fragmentation and cleanup:
+ Mix the following on ice:
+aRNA 16μl
+Fragmentation buffer 4μl
+ Incubate for 3 min. at 94 o C.
+ Immediately move to ice and add 2μl fragmentation stop buffer.
+ Adjust volume to 30ul by adding 8μl water.
+ Add 105μl aRNA binding buffer, followed by 75μl EtOH, immediately mix by pipetting 3-4 times
+and load onto spin-column. Bind sample by immediately spinning for 1 min at 10,000g, discard
+flow through. (If cleaning more than one reaction, this step should be done for each sample
+separately)
+ Add 500μl wash buffer (samples can wait at this step until binding of all samples is complete).
+ Spin as above, discard flow through.
+ Spin for an additional minute to dry.
+ Transfer column to clean tube.
+ Elute by adding 10μl warm water (55 o C), incubating for 2 minutes at room temperature and
+spinning 1.5 min at 10,000 g.
+ Repeat elution.
+Stopping point: Samples can be kept at -80 o C
+Check aRNA amount and quality:
+ Load 1μl onto Bioanalyzer RNA pico chip after heating an aliquot of the sample to 70 o for 2 min.
+ When starting the IVT with ~0.5ng total RNA, the expected yield is 500-1000 pg/μl. Size
+distribution should peak at ~500 bp (See Bioanalyzer plot for example).
+
+### Page 6
+
+Library preparation:
+Protocol designed for 5-10ng amplified RNA, although as little as 1-2ng can be used, but then additional
+PCR cycles are required. Sample volume should be adjusted to 16μl, either by adding water or drying
+down in a speedvac, depending on RNA concentration. IVTs can be pooled at this point if there is no
+overlap in barcodes used.
+Phosphatase treatment:
+To 16μl of fragmented aRNA in a 0.7ml PCR tube add 4μl of the following mix:
+ 10X phospatase buffer 2μl
+ Antarctic phosphatase 1μl
+ RNaseOUT 1μl
+Incubate in a thermal cycler with the following protocol:
+ 37°C for 30 minutes
+ 65°C for 5 minutes
+ 4°C indefinite hold
+PNK treatment:
+To the 0.7 ml PCR tube from the previous step add 30μl of the following mix:
+ nuclease-free H2O 17μl
+ 10X phosphatase buffer 5μl
+ ATP (10mM, from Illumina kit) 5μl
+ RNaseOUT 1μl
+ PNK 2μl
+Incubate in a thermal cycler at 37°C for 60 minutes then 4°C hold.
+Column cleanup of phosphatase and PNK treated aRNA (RNeasy kit):
+ Adjust the volume to 100μl (add 50μL) using nuclease free water.
+ Add 350μl of RLT buffer and mix well.
+ Add 250μl EtOH, mix well by pipetting, and transfer sample to an RNeasy spin column.
+ Spin 15 sec. at 8,000 g.
+ Transfer column to new collection tube, and add 500μl Buffer RPE.
+ Spin 15 sec. at 8,000 g.
+ Discard flow-through, and add 500 ml 80% EtOH.
+ Spin 2 min. at 8,000 g.
+ Transfer column to new collection tube, open lid of column, and spin 5 min at full speed.
+ Transfer column to new collection tube, and elute with 14μl nuclease free H2O, spinning at full
+speed for 1 minute.
+ Dry down the sample using a speedvac to 5μl. (approx. 7 minutes)
+
+### Page 7
+
+Ligate 3' adapter:
+Dilute 3' adapter (RA3, from Illumina kit) 5 fold.
+To 5μl phosphatase and PNK treated RNA add 1μl of the diluted 3' adaptor.
+Incubate at 70°C for 2 minutes and then immediately place the tube on ice to prevent secondary
+structure formation.
+Add 4μl of the following mix:
+ 5X HM Ligation Buffer (HML, Illumina kit) 2μL
+ RNase Inhibitor (Illumina kit) 1μL
+ T4 RNA Ligase 2, truncated 1μL
+Incubate the tube on the pre-heated thermal cycler at 28°C for 1 hour (with unheated or open lid).
+With the reaction tube remaining on the thermal cycler, add 1μl Stop Solution (STP, Illumina kit) and
+gently pipette the entire volume up and down 6-8 times to mix thoroughly. Continue to incubate the
+reaction tube on the thermal cycler at 28°C for 15 minutes, and then place the tube on ice.
+Add 3μL nuclease free water.
+Reverse transcription reaction:
+Dilute dNTPs (from Illumina kit) two fold with nuclease free water (prepare at least 1μl per sample)
+Combine the following in a PCR tube (the remaining 3' adapter-ligated RNA may be stored at -80°C):
+ Adapter-ligated RNA 6μL
+ RNA RT Primer (RTP, from Illumina kit)   1μL
+Incubate the tube at 70°C for 2 minutes and then immediately place the tube on ice.
+Add 5.5μl of the following mix:
+ 5X First Strand Buffer 2μL
+ 12.5 mM dNTP mix (diluted dNTP) 0.5μL
+ 100 mM DTT 1μL
+ RNase Inhibitor (Illumina kit) 1μL
+ SuperScript II Reverse Transcriptase 1μL
+Incubate the tube in the pre-heated thermal cycler at 50°C for 1 hour and then place the tube on ice.
+
+### Page 8
+
+PCR amplification:
+To each reverse transcription reaction add 35.5μl of the following mix:
+ Ultra Pure Water 8.5μL
+ PCR mix (PML, from Illumina kit)) 25μL
+ RNA PCR Primer (RP1, from Illumina kit) 2μL
+To each reaction add 2μl of a uniquely indexed RNA PCR Primer (RPIX, from Illumina kit)
+Amplify the tube in the thermal cycler using the following PCR cycling conditions:
+ 30 seconds at 98°C
+ 12 cycles of:
+ 10 seconds at 98°C
+ 30 seconds at 60°C
+ 30 seconds at 72°C
+ 10 minutes at 72°C
+ Hold at 4°C
+Can go up to 15 cycles if necessary, or down to 11 if starting with the full 10ng.
+Stopping point: samples can be kept at -20 o C.
+Bead Cleanup of PCR products - Repeat 1:
+ Prewarm beads to room temperature.
+ Vortex AMPure XP Beads until well dispersed, then add 50μl to the 50μl PCR reaction. Mix entire
+volume up ten times to mix thoroughly.
+ Incubate at room temperature for 15 min.
+ Place on magnetic stand for at least 5 min, until liquid appears clear.
+ Remove and discard 95μl of the supernatant.
+ Add 200μl freshly prepared 80% EtOH.
+ Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+ Add 200μl freshly prepared 80% EtOH
+ Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+ Air dry beads for 15 min, or until completely dry.
+ Resuspend with 32.5μl Resuspension Buffer (from Illumina kit). Pipette entire volume up and
+down ten times to mix thoroughly.
+ Incubate at room temperature for 2 min.
+ Place on magnetic stand for 5 min, until liquid appears clear.
+ Transfer 30μl of supernatant to new tube.
+
+### Page 9
+
+Bead Cleanup of PCR products - Repeat 2:
+Repeat as above, but adding 39μl beads and eluting in 12.5μl resuspension buffer at the end,
+transferring 10μl to a new tube.
+Check library amount and quality:
+Check concentration of DNA by Qubit, 1μl should be enough to measure using the high sensitivity
+reagent; expected concentration is at least ~1ng/μl.
+Run 1μl of each sample on Bioanalyzer using a high sensitivity DNA chip to see size distribution.
+Expected peak at 300-400bp (See Bioanalyzer plot for example).
+Concentration to be loaded for sequencing should be calibrated by the sequencing facility. For us, 5pM
+on Hi-Seq v.1 reagents gave good cluster density.

cel_seq/CEL-Seq_protocol.pymupdf_text.txt

@@ -0,0 +1,391 @@
+# pymupdf text extraction
+source_pdf: CEL-Seq_protocol.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/cel_seq/CEL-Seq_protocol.pdf
+extraction: PyMuPDF: page.get_text("text", sort=True)
+
+## Page 1
+
+CEL-Seq Protocol
+By Tamar Hashimshony, Technion
+July 16th, 2012
+
+
+Reagents:
+LoBind tubes – 0.5 ml – Eppendorf 022431005
+Ultra pure RNase free water
+Ethanol
+Bioanalyzer kits - Agilent RNA pico kit (5067-1513), high sensitivity DNA kit (5067-4626)
+Qubit reagents: dsDNA HS Assay – invitrogen Q32851 or Q32854
+For RNA amplification:
+ERCC RNA spike-in mix – Ambion 4456740
+MessageAmpII kit – Ambion AM1751
+Optional: extra columns for cDNA/aRNA purification – Ambion10066G
+Fragmentation buffer: 200mM Tris-acetate, pH 8.1, 500 mM KOAc, 150 mM MgOAc
+Fragmentation stop buffer: 0.5 M EDTA
+For Library preparation:
+Antarctic phosphatase – NEB M0289
+RNase OUT – invitrogen 100000840
+PNK – NEB M0201
+Superscript II – invitrogen 18064-014
+RNeasy MinElute kit – Qiagen 74204
+T4 RNA ligase 2, truncated – NEB M0242
+AMPure XP beads – Beckman Coulter A63880
+TruSeq small RNA sample prep kit – Illumina RS-200-0012 (or -0024, -0036, -0048)
+Optional (to supplement Illumina’s Small RNA kit):
+ATP – NEB P0756
+Phusion® High-Fidelity PCR Master Mix with HF Buffer – NEB M0531
+RNA RT primer, RNA PCR primers (sequences available from Illumina)
+
+
+Equipment:
+Thermocycler with lid with adjustable temperature (one that can also fit 0.5 ml PCR tubes is convenient)
+Speed vac
+Oven
+Heat block
+Magnetic stand (for 0.5 ml tubes)
+Qubit® Fluorometer - invitrogen
+Bioanalyzer – Agilent
+
+## Page 2
+
+Primers:
+CEL-Seq primer design: The RT primer was designed with an anchored polyT, a unique barcode, the 5’
+Illumina adapter (as used in the Illumina small RNA kit) and a T7 promoter. The barcodes were of length
+eight and designed in groups of four, such that the first five nucleotides will have equal representation
+of all four nucleotides to allow for template generation and crosstalk corrections which are based on the
+first four nucleotides read in the Illumina platform. The barcodes were designed such that each pair is
+different by at least two nucleotides, so that a single sequencing error will not produce the wrong
+barcode. Primers are desalted, stock solution 1 μg/μl, working concentration 25ng/μl.
+
+
+#1: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#2: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#3: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#4: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#5: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#6: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#7: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#8: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#9: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCATCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#10: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#11: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#12: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#13: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#14: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#15: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#16: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#17: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#18: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#19: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#20: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#21: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#22: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#23: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#24: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#25: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#26: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#27:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#28: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#29: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#30: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#31: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#32: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#33: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#34: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#35: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#36: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#37: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#38: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#39:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#40: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+## Page 3
+
+Single cell isolation:
+
+Individual cells (so far we’ve worked with C. elegans blastomeres or trypsinised tissue culture cells) are
+transferred with a micro-pipette into a 0.5µl drop of appropriate buffer (egg salts or PBS) placed on the
+cap of a 0.5 ml LoBind Eppendorf tube. Location of cell should be marked. Excess liquid is aspirated off,
+and tube is frozen in liquid nitrogen. Samples are stored at -80°C.
+
+
+
+
+RNA Amplification:
+
+Prepare primer mix (for each different primer used):
+
+    Primer (25ng/μl)   1μl
+    ERCC Spike-in     Xμl
+    water             Yμl
+                        6μl
+
+Spike-in dilution should be appropriate for sample size – see protocol of ERCC RNA spike in mix. For
+single cells we add 1ul of spike-in at 1:500,000 dilution.
+
+Breaking cell open and annealing with primer:
+
+      Add 1.2μl primer mix to marked location of single cell on cap of tube (Keep cell frozen until
+       adding the primer mix, handle up to 12 cells in parallel).
+       Incubate 5 min. at 70oC (with lid of thermal cycler set to 70oC).
+       Brief spin down.
+       Incubate for an additional 5 min. at 70oC.
+      Move immediately to ice.
+       Spin at maximal speed for a few seconds to collect as many droplets as possible before next
+        step, and then return to ice.
+
+RT reaction (Ambion kit)
+
+      Add 0.8μl of the following mix to each reaction:
+                   First Strand buffer      0.2μl
+            dNTP                   0.4μl
+             RNase Inhibitor         0.1μl
+                ArrayScript             0.1μl
+       Incubate 2hr at 42oC ( in hybridization oven)
+
+Second strand reaction (Ambion kit):
+
+      Move previous step to ice so it cools below 16oC.
+      Add 8uL of the following mix to each reaction tube:
+
+## Page 4
+
+         DDW                   6.3μl
+             Second strand buf.     1μl
+            dNTP                   0.4μl
+           DNA Pol                0.2μl
+            RNaseH                 0.1μl
+
+         Flick and spin samples (at maximal speed for a few seconds).
+
+       Incubate at 16oC for 2hr (in thermal cycler with unheated or open lid).
+
+cDNA cleanup and speedvac:
+
+       Pool all cells that are to go to same IVT. Should have ~10μl from each cell.
+       Adjust volume to 100μl with nuclease free water.  If more than 10 cells in a pool, just add all
+        together.
+      Add 250μl cDNA binding buffer to each 100ul sample (if total sample volume exceeds 100uL,
+        adjust cDNA binding buffer volume according to sample volume). Load onto Ambion cDNA
+       cleanup column. Volume of up to 24 pooled samples can be loaded. If more than 24 samples are
+       pooled, after spin load remaining volume and spin again.
+       Spin 1 min at 10,000g to bind cDNA, discard flow through (repeat if more than 24 samples are
+        pooled).
+      Add 500μl wash buffer, spin as above, discard flow through.
+       Spin for an additional minute to dry.
+       Transfer column to clean round bottom 2 ml tube.
+       Elute by adding 9μl warm water (55oC), incubating for 2 minutes at room temperature and
+        spinning 1.5 min at 10,000g.
+      Repeat elution.
+       Adjust volume to 6.4μl by drying in a speedvac (~8 min).
+
+Stopping point: Samples can be kept at -20oC
+
+IVT (Ambion kit):
+
+       Prepare the following mix and add 9.6μl per tube.
+           A                       1.6μl
+          G                       1.6μl
+            C                       1.6μl
+          U                       1.6μl
+             10xT7 buffer            1.6μl
+             T7 enzyme              1.6μl
+       Incubate in a thermal cycler at 37oC for 13 hrs, with lid at 70oC. Set cycler to go to 4oC at end of
+        incubation. aRNA (amplified RNA) is stable for at least several hours.
+
+## Page 5
+
+RNA fragmentation and cleanup:
+
+      Mix the following on ice:
+            aRNA                 16μl
+              Fragmentation buffer   4μl
+       Incubate for 3 min. at 94oC.
+       Immediately move to ice and add 2μl fragmentation stop buffer.
+       Adjust volume to 30ul by adding 8μl water.
+      Add 105μl aRNA binding buffer, followed by 75μl EtOH, immediately mix by pipetting 3-4 times
+      and load onto spin-column. Bind sample by immediately spinning for 1 min at 10,000g, discard
+       flow through. (If cleaning more than one reaction, this step should be done for each sample
+        separately)
+      Add 500μl wash buffer (samples can wait at this step until binding of all samples is complete).
+       Spin as above, discard flow through.
+       Spin for an additional minute to dry.
+       Transfer column to clean tube.
+       Elute by adding 10μl warm water (55oC), incubating for 2 minutes at room temperature and
+        spinning 1.5 min at 10,000 g.
+      Repeat elution.
+
+Stopping point: Samples can be kept at -80oC
+
+Check aRNA amount and quality:
+
+      Load 1μl onto Bioanalyzer RNA pico chip after heating an aliquot of the sample to 70o for 2 min.
+      When starting the IVT with ~0.5ng total RNA, the expected yield is 500-1000 pg/μl. Size
+        distribution should peak at ~500 bp (See Bioanalyzer plot for example).
+
+## Page 6
+
+Library preparation:
+
+Protocol designed for 5-10ng amplified RNA, although as little as 1-2ng can be used, but then additional
+PCR cycles are required. Sample volume should be adjusted to 16μl, either by adding water or drying
+down in a speedvac, depending on RNA concentration. IVTs can be pooled at this point if there is no
+overlap in barcodes used.
+
+Phosphatase treatment:
+To 16μl of fragmented aRNA in a 0.7ml PCR tube add 4μl of the following mix:
+      10X phospatase buffer         2μl
+       Antarctic phosphatase         1μl
+      RNaseOUT                    1μl
+
+
+Incubate in a thermal cycler with the following protocol:
+      37°C for 30 minutes
+      65°C for 5 minutes
+       4°C indefinite hold
+
+
+PNK treatment:
+To the 0.7 ml PCR tube from the previous step add 30μl of the following mix:
+       nuclease-free H2O            17μl
+      10X phosphatase buffer        5μl
+      ATP (10mM, from Illumina kit)  5μl
+      RNaseOUT                    1μl
+      PNK                          2μl
+Incubate in a thermal cycler at 37°C for 60 minutes then 4°C hold.
+
+
+
+Column cleanup of phosphatase and PNK treated aRNA (RNeasy kit):
+       Adjust the volume to 100μl (add 50μL) using nuclease free water.
+      Add 350μl of RLT buffer and mix well.
+      Add 250μl EtOH, mix well by pipetting, and transfer sample to an RNeasy spin column.
+       Spin 15 sec. at 8,000 g.
+       Transfer column to new collection tube, and add 500μl Buffer RPE.
+       Spin 15 sec. at 8,000 g.
+       Discard flow-through, and add 500 ml 80% EtOH.
+       Spin 2 min. at 8,000 g.
+       Transfer column to new collection tube, open lid of column, and spin 5 min at full speed.
+       Transfer column to new collection tube, and elute with 14μl nuclease free H2O, spinning at full
+       speed for 1 minute.
+      Dry down the sample using a speedvac to 5μl. (approx. 7 minutes)
+
+## Page 7
+
+Ligate 3’ adapter:
+
+Dilute 3’ adapter (RA3, from Illumina kit) 5 fold.
+
+To 5μl phosphatase and PNK treated RNA add 1μl of the diluted 3’ adaptor.
+
+Incubate at 70°C for 2 minutes and then immediately place the tube on ice to prevent secondary
+structure formation.
+
+Add 4μl of the following mix:
+
+      5X HM Ligation Buffer (HML, Illumina kit)      2μL
+      RNase Inhibitor (Illumina kit)                 1μL
+      T4 RNA Ligase 2, truncated                  1μL
+
+Incubate the tube on the pre‐heated thermal cycler at 28°C for 1 hour (with unheated or open lid).
+
+With the reaction tube remaining on the thermal cycler, add 1μl Stop Solution (STP, Illumina kit) and
+gently pipette the entire volume up and down 6–8 times to mix thoroughly. Continue to incubate the
+reaction tube on the thermal cycler at 28°C for 15 minutes, and then place the tube on ice.
+
+Add 3μL nuclease free water.
+
+
+
+
+Reverse transcription reaction:
+
+Dilute dNTPs (from Illumina kit) two fold with nuclease free water (prepare at least 1μl per sample)
+
+Combine the following in a PCR tube (the remaining 3’ adapter‐ligated RNA may be stored at ‐80°C):
+
+       Adapter‐ligated RNA                6μL
+      RNA RT Primer (RTP, from Illumina kit)  1μL
+
+Incubate the tube at 70°C for 2 minutes and then immediately place the tube on ice.
+
+Add 5.5μl of the following mix:
+
+      5X First Strand Buffer                2μL
+       12.5 mM dNTP mix (diluted dNTP)      0.5μL
+      100 mM DTT                       1μL
+      RNase Inhibitor (Illumina kit)          1μL
+       SuperScript II Reverse Transcriptase    1μL
+
+Incubate the tube in the pre‐heated thermal cycler at 50°C for 1 hour and then place the tube on ice.
+
+## Page 8
+
+PCR amplification:
+
+To each reverse transcription reaction add 35.5μl of the following mix:
+
+       Ultra Pure Water                            8.5μL
+      PCR mix (PML, from Illumina kit))            25μL
+      RNA PCR Primer (RP1, from Illumina kit)       2μL
+
+To each reaction add 2μl of a uniquely indexed RNA PCR Primer (RPIX, from Illumina kit)
+
+Amplify the tube in the thermal cycler using the following PCR cycling conditions:
+      30 seconds at 98°C
+      12 cycles of:
+          10 seconds at 98°C
+          30 seconds at 60°C
+          30 seconds at 72°C
+      10 minutes at 72°C
+      Hold at 4°C
+Can go up to 15 cycles if necessary, or down to 11 if starting with the full 10ng.
+
+Stopping point: samples can be kept at -20oC.
+
+
+
+
+Bead Cleanup of PCR products – Repeat 1:
+
+      Prewarm beads to room temperature.
+       Vortex AMPure XP Beads until well dispersed, then add 50μl to the 50μl PCR reaction. Mix entire
+      volume up ten times to mix thoroughly.
+       Incubate at room temperature for 15 min.
+       Place on magnetic stand for at least 5 min, until liquid appears clear.
+      Remove and discard 95μl of the supernatant.
+      Add 200μl freshly prepared 80% EtOH.
+       Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+      Add 200μl freshly prepared 80% EtOH
+       Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+       Air dry beads for 15 min, or until completely dry.
+      Resuspend with 32.5μl Resuspension Buffer (from Illumina kit). Pipette entire volume up and
+      down ten times to mix thoroughly.
+       Incubate at room temperature for 2 min.
+       Place on magnetic stand for 5 min, until liquid appears clear.
+       Transfer 30μl of supernatant to new tube.
+
+## Page 9
+
+Bead Cleanup of PCR products – Repeat 2:
+
+Repeat as above, but adding 39μl beads and eluting in 12.5μl resuspension buffer at the end,
+transferring 10μl to a new tube.
+
+
+
+
+Check library amount and quality:
+
+Check concentration of DNA by Qubit, 1μl should be enough to measure using the high sensitivity
+reagent; expected concentration is at least ~1ng/μl.
+
+Run 1μl of each sample on Bioanalyzer using a high sensitivity DNA chip to see size distribution.
+Expected peak at 300-400bp (See Bioanalyzer plot for example).
+
+
+
+
+
+Concentration to be loaded for sequencing should be calibrated by the sequencing facility. For us, 5pM
+on Hi-Seq v.1 reagents gave good cluster density.

cel_seq/CEL-Seq_protocol.pypdf_text.txt

@@ -0,0 +1,396 @@
+# pypdf text extraction
+source_pdf: CEL-Seq_protocol.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/cel_seq/CEL-Seq_protocol.pdf
+extraction: pypdf: page.extract_text(extraction_mode="layout")
+
+## Page 1
+
+CEL-Seq Protocol
+By Tamar Hashimshony, Technion
+July 16th, 2012
+
+
+Reagents:
+LoBind tubes – 0.5 ml – Eppendorf 022431005
+Ultra pure RNase free water
+Ethanol
+Bioanalyzer kits - Agilent RNA pico kit (5067-1513), high sensitivity DNA kit (5067-4626)
+Qubit reagents: dsDNA HS Assay – invitrogen Q32851 or Q32854
+For RNA amplification:
+ERCC RNA spike-in mix – Ambion 4456740
+MessageAmpII kit – Ambion AM1751
+Optional: extra columns for cDNA/aRNA purification – Ambion10066G
+Fragmentation buffer: 200mM Tris-acetate, pH 8.1, 500 mM KOAc, 150 mM MgOAc
+Fragmentation stop buffer: 0.5 M EDTA
+For Library preparation:
+Antarctic phosphatase – NEB M0289
+RNase OUT – invitrogen 100000840
+PNK – NEB M0201
+Superscript II – invitrogen 18064-014
+RNeasy MinElute kit – Qiagen 74204
+T4 RNA ligase 2, truncated – NEB M0242
+AMPure XP beads – Beckman Coulter A63880
+TruSeq small RNA sample prep kit – Illumina RS-200-0012 (or -0024, -0036, -0048)
+Optional (to supplement Illumina’s Small RNA kit):
+ATP – NEB P0756
+Phusion® High-Fidelity PCR Master Mix with HF Buffer – NEB M0531
+RNA RT primer, RNA PCR primers (sequences available from Illumina)
+
+Equipment:
+Thermocycler with lid with adjustable temperature (one that can also fit 0.5 ml PCR tubes is convenient)
+Speed vac
+Oven
+Heat block
+Magnetic stand (for 0.5 ml tubes)
+Qubit® Fluorometer - invitrogen
+Bioanalyzer – Agilent
+
+## Page 2
+
+Primers:
+CEL-Seq primer design: The RT primer was designed with an anchored polyT, a unique barcode, the 5’
+Illumina adapter (as used in the Illumina small RNA kit) and a T7 promoter. The barcodes were of length
+eight and designed in groups of four, such that the first five nucleotides will have equal representation
+of all four nucleotides to allow for template generation and crosstalk corrections which are based on the
+first four nucleotides read in the Illumina platform. The barcodes were designed such that each pair is
+different by at least two nucleotides, so that a single sequencing error will not produce the wrong
+barcode. Primers are desalted, stock solution 1 μg/μl, working concentration 25ng/μl.
+
+#1:  CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#2: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#3: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#4: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#5: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#6: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#7: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#8: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGATCTTTTTTTTTTTTTTTTTTTTTTTTV
+#9: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCATCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#10: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#11: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#12: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGTCCTTTTTTTTTTTTTTTTTTTTTTTTV
+#13: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCATCAGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#14: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGTCGTGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#15: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCACGACGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#16: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTGATGGAATTTTTTTTTTTTTTTTTTTTTTTTV
+#17: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#18: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#19: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#20: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#21: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#22: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#23: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#24: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#25: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTCACAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#26: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCGTGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#27:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGACACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#28: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCATGTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#29: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#30: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#31: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCACGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#32: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTCGCTTTTTTTTTTTTTTTTTTTTTTTTV
+#33: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#34: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#35: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCATCATTTTTTTTTTTTTTTTTTTTTTTTV
+#36: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTTCATTTTTTTTTTTTTTTTTTTTTTTTV
+#37: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCCTAACGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#38: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCGCTTGGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#39:CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCAGCCAGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+#40: CGATTGAGGCCGGTAATACGACTCACTATAGGGGTTCAGAGTTCTACAGTCCGACGATCTAGGTGAGTTTTTTTTTTTTTTTTTTTTTTTTV
+
+## Page 3
+
+Single cell isolation:
+
+Individual cells (so far we’ve worked with C. elegans blastomeres or trypsinised tissue culture cells) are
+transferred with a micro-pipette into a 0.5µl drop of appropriate buffer (egg salts or PBS) placed on the
+cap of a 0.5 ml LoBind Eppendorf tube. Location of cell should be marked. Excess liquid is aspirated off,
+and tube is frozen in liquid nitrogen. Samples are stored at -80°C.
+
+
+
+RNA Amplification:
+
+Prepare primer mix (for each different primer used):
+
+      Primer (25ng/μl)         1μl
+      ERCC Spike-in          Xμl
+      water                    Yμl
+                               6μl
+
+Spike-in dilution should be appropriate for sample size – see protocol of ERCC RNA spike in mix. For
+single cells we add 1ul of spike-in at 1:500,000 dilution.
+
+Breaking cell open and annealing with primer:
+
+         Add 1.2μl primer mix to marked location of single cell on cap of tube (Keep cell frozen until
+          adding the primer mix, handle up to 12 cells in parallel).
+         Incubate 5 min. at 70oC (with lid of thermal cycler set to 70oC).
+         Brief spin down.
+         Incubate for an additional 5 min. at 70oC.
+         Move immediately to ice.
+         Spin at maximal speed for a few seconds to collect as many droplets as possible before next
+          step, and then return to ice.
+
+RT reaction (Ambion kit)
+
+         Add 0.8μl of the following mix to each reaction:
+                    First Strand buffer            0.2μl
+                    dNTP                           0.4μl
+                    RNase Inhibitor                0.1μl
+                    ArrayScript                    0.1μl
+         Incubate 2hr at 42oC ( in hybridization oven)
+
+Second strand reaction (Ambion kit):
+         Move previous step to ice so it cools below 16oC.
+         Add 8uL of the following mix to each reaction tube:
+
+## Page 4
+
+                    DDW                           6.3μl
+                    Second strand buf.            1μl
+                    dNTP                          0.4μl
+                    DNA Pol                       0.2μl
+                    RNaseH                        0.1μl
+
+          Flick and spin samples (at maximal speed for a few seconds).
+         Incubate at 16oC for 2hr (in thermal cycler with unheated or open lid).
+
+cDNA cleanup and speedvac:
+
+         Pool all cells that are to go to same IVT.  Should have ~10μl from each cell.
+         Adjust volume to 100μl with nuclease free water.  If more than 10 cells in a pool, just add all
+          together.
+         Add 250μl cDNA binding buffer to each 100ul sample (if total sample volume exceeds 100uL,
+          adjust cDNA binding buffer volume according to sample volume). Load onto Ambion cDNA
+          cleanup column. Volume of up to 24 pooled samples can be loaded. If more than 24 samples are
+          pooled, after spin load remaining volume and spin again.
+         Spin 1 min at 10,000g to bind cDNA, discard flow through (repeat if more than 24 samples are
+          pooled).
+         Add 500μl wash buffer, spin as above, discard flow through.
+         Spin for an additional minute to dry.
+         Transfer column to clean round bottom 2 ml tube.
+         Elute by adding 9μl warm water (55oC), incubating for 2 minutes at room temperature and
+          spinning 1.5 min at 10,000g.
+         Repeat elution.
+         Adjust volume to 6.4μl by drying in a speedvac (~8 min).
+Stopping point: Samples can be kept at -20oC
+
+IVT (Ambion kit):
+
+         Prepare the following mix and add 9.6μl per tube.
+                    A                             1.6μl
+                    G                             1.6μl
+                    C                             1.6μl
+                    U                             1.6μl
+                    10xT7 buffer                  1.6μl
+                    T7 enzyme                     1.6μl
+         Incubate in a thermal cycler at 37oC for 13 hrs, with lid at 70oC.  Set cycler to go to 4oC at end of
+          incubation. aRNA (amplified RNA) is stable for at least several hours.
+
+## Page 5
+
+RNA fragmentation and cleanup:
+
+         Mix the following on ice:
+                    aRNA                          16μl
+                    Fragmentation buffer          4μl
+         Incubate for 3 min. at 94oC.
+         Immediately move to ice and add 2μl fragmentation stop buffer.
+         Adjust volume to 30ul by adding 8μl water.
+         Add 105μl aRNA binding buffer, followed by 75μl EtOH, immediately mix by pipetting 3-4 times
+          and load onto spin-column. Bind sample by immediately spinning for 1 min at 10,000g, discard
+          flow through. (If cleaning more than one reaction, this step should be done for each sample
+          separately)
+         Add 500μl wash buffer (samples can wait at this step until binding of all samples is complete).
+         Spin as above, discard flow through.
+         Spin for an additional minute to dry.
+         Transfer column to clean tube.
+         Elute by adding 10μl warm water (55oC), incubating for 2 minutes at room temperature and
+          spinning 1.5 min at 10,000 g.
+         Repeat elution.
+Stopping point: Samples can be kept at -80oC
+
+Check aRNA amount and quality:
+         Load 1μl onto Bioanalyzer RNA pico chip after heating an aliquot of the sample to 70o for 2 min.
+         When starting the IVT with ~0.5ng total RNA, the expected yield is 500-1000 pg/μl. Size
+          distribution should peak at ~500 bp (See Bioanalyzer plot for example).
+
+## Page 6
+
+Library preparation:
+
+Protocol designed for 5-10ng amplified RNA, although as little as 1-2ng can be used, but then additional
+PCR cycles are required. Sample volume should be adjusted to 16μl, either by adding water or drying
+down in a speedvac, depending on RNA concentration. IVTs can be pooled at this point if there is no
+overlap in barcodes used.
+
+Phosphatase treatment:
+To 16μl of fragmented aRNA in a 0.7ml PCR tube add 4μl of the following mix:
+         10X phospatase buffer                   2μl
+         Antarctic phosphatase                   1μl
+         RNaseOUT                                1μl
+
+Incubate in a thermal cycler with the following protocol:
+         37°C for 30 minutes
+         65°C for 5 minutes
+         4°C indefinite hold
+
+PNK treatment:
+To the 0.7 ml PCR tube from the previous step add 30μl of the following mix:
+         nuclease-free H2O                       17μl
+         10X phosphatase buffer                  5μl
+         ATP (10mM, from Illumina kit)           5μl
+         RNaseOUT                                1μl
+         PNK                                     2μl
+Incubate in a thermal cycler at 37°C for 60 minutes then 4°C hold.
+
+
+Column cleanup of phosphatase and PNK treated aRNA (RNeasy kit):
+         Adjust the volume to 100μl (add 50μL) using nuclease free water.
+         Add 350μl of RLT buffer and mix well.
+         Add 250μl EtOH, mix well by pipetting, and transfer sample to an RNeasy spin column.
+         Spin 15 sec. at 8,000 g.
+         Transfer column to new collection tube, and add 500μl Buffer RPE.
+         Spin 15 sec. at 8,000 g.
+         Discard flow-through, and add 500 ml 80% EtOH.
+         Spin 2 min. at 8,000 g.
+         Transfer column to new collection tube, open lid of column, and spin 5 min at full speed.
+         Transfer column to new collection tube, and elute with 14μl nuclease free H2O, spinning at full
+          speed for 1 minute.
+         Dry down the sample using a speedvac to 5μl. (approx. 7 minutes)
+
+## Page 7
+
+Ligate 3’ adapter:
+
+Dilute 3’ adapter (RA3, from Illumina kit) 5 fold.
+
+To 5μl phosphatase and PNK treated RNA add 1μl of the diluted 3’ adaptor.
+
+Incubate at 70°C for 2 minutes and then immediately place the tube on ice to prevent secondary
+structure formation.
+
+Add 4μl of the following mix:
+
+         5X HM Ligation Buffer (HML, Illumina kit)                   2μL
+         RNase Inhibitor (Illumina kit)                              1μL
+         T4 RNA Ligase 2, truncated                                  1μL
+
+Incubate the tube on the pre‐heated thermal cycler at 28°C for 1 hour (with unheated or open lid).
+
+With the reaction tube remaining on the thermal cycler, add 1μl Stop Solution (STP, Illumina kit) and
+gently pipette the entire volume up and down 6–8 times to mix thoroughly. Continue to incubate the
+reaction tube on the thermal cycler at 28°C for 15 minutes, and then place the tube on ice.
+
+Add 3μL nuclease free water.
+
+
+
+Reverse transcription reaction:
+
+Dilute dNTPs (from Illumina kit) two fold with nuclease free water (prepare at least 1μl per sample)
+
+Combine the following in a PCR tube (the remaining 3’ adapter‐ligated RNA may be stored at ‐80°C):
+
+         Adapter‐ligated RNA                               6μL
+         RNA RT Primer (RTP, from Illumina kit)            1μL
+
+Incubate the tube at 70°C for 2 minutes and then immediately place the tube on ice.
+
+Add 5.5μl of the following mix:
+
+         5X First Strand Buffer                            2μL
+         12.5 mM dNTP mix (diluted dNTP)                   0.5μL
+         100 mM DTT                                        1μL
+         RNase Inhibitor (Illumina kit)                    1μL
+         SuperScript II Reverse Transcriptase              1μL
+
+Incubate the tube in the pre‐heated thermal cycler at 50°C for 1 hour and then place the tube on ice.
+
+## Page 8
+
+PCR amplification:
+
+To each reverse transcription reaction add 35.5μl of the following mix:
+
+         Ultra Pure Water                                            8.5μL
+         PCR mix (PML, from Illumina kit))                           25μL
+         RNA PCR Primer (RP1, from Illumina kit)                     2μL
+
+To each reaction add 2μl of a uniquely indexed RNA PCR Primer (RPIX, from Illumina kit)
+
+Amplify the tube in the thermal cycler using the following PCR cycling conditions:
+         30 seconds at 98°C
+         12 cycles of:
+               10 seconds at 98°C
+               30 seconds at 60°C
+               30 seconds at 72°C
+         10 minutes at 72°C
+         Hold at 4°C
+Can go up to 15 cycles if necessary, or down to 11 if starting with the full 10ng.
+Stopping point: samples can be kept at -20oC.
+
+
+
+Bead Cleanup of PCR products – Repeat 1:
+
+         Prewarm beads to room temperature.
+         Vortex AMPure XP Beads until well dispersed, then add 50μl to the 50μl PCR reaction. Mix entire
+          volume up ten times to mix thoroughly.
+         Incubate at room temperature for 15 min.
+         Place on magnetic stand for at least 5 min, until liquid appears clear.
+         Remove and discard 95μl of the supernatant.
+         Add 200μl freshly prepared 80% EtOH.
+         Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+         Add 200μl freshly prepared 80% EtOH
+         Incubate at least 30 seconds, then remove and discard supernatant without disturbing beads.
+         Air dry beads for 15 min, or until completely dry.
+         Resuspend with 32.5μl Resuspension Buffer (from Illumina kit). Pipette entire volume up and
+          down ten times to mix thoroughly.
+         Incubate at room temperature for 2 min.
+         Place on magnetic stand for 5 min, until liquid appears clear.
+         Transfer 30μl of supernatant to new tube.
+
+## Page 9
+
+Bead Cleanup of PCR products – Repeat 2:
+
+Repeat as above, but adding 39μl beads and eluting in 12.5μl resuspension buffer at the end,
+transferring 10μl to a new tube.
+
+
+
+Check library amount and quality:
+
+Check concentration of DNA by Qubit, 1μl should be enough to measure using the high sensitivity
+reagent; expected concentration is at least ~1ng/μl.
+
+Run 1μl of each sample on Bioanalyzer using a high sensitivity DNA chip to see size distribution.
+Expected peak at 300-400bp (See Bioanalyzer plot for example).
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Concentration to be loaded for sequencing should be calibrated by the sequencing facility. For us, 5pM
+on Hi-Seq v.1 reagents gave good cluster density.

drop_seq/drop-seq.human_text.txt

@@ -0,0 +1,259 @@
+Drop-seq
+
+Protocol
+
+Drop-seq profiles single-cell RNA expression by co-encapsulating single cells and barcoded primer beads in nanoliter droplets. One aqueous flow contains cells, and the other contains barcoded primer beads suspended in lysis buffer. After droplet formation, cells lyse inside droplets and released mRNAs hybridize to bead-bound barcoded oligo-dT primers. Droplets are then broken, beads are collected and washed, and mRNAs attached to beads are reverse transcribed in bulk to form STAMPs, single-cell transcriptomes attached to microparticles. After exonuclease I treatment, bead aliquots are PCR amplified. The amplified cDNA is then prepared for sequencing using Nextera XT tagmentation with custom primers.
+
+The bead-bound primers contain a common PCR handle, a cell barcode, a UMI, and an oligo-dT capture region. The cell barcode identifies the bead/cell of origin, and the UMI identifies individual captured mRNA molecules. Drop-seq libraries are sequenced paired-end. Read 1 is 20 bp: bases 1–12 are the cell barcode and bases 13–20 are the UMI. Read 2 sequences the cDNA insert and is used to identify the gene of origin.
+
+Key oligo and library-related sequences
+
+The explicit oligo sequences are provided in Table S6 of the supplemental information.
+
+1. Barcoded Bead SeqA
+
+Source sequence:
+
+5'-Bead-Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACGTJJJJJJJJJJJJNNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3'
+
+2. Barcoded Bead SeqB
+
+Source sequence:
+
+5'-Bead-Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACJJJJJJJJJJJJNNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3'
+
+3. Template_Switch_Oligo
+
+Source sequence:
+
+AAGCAGTGGTATCAACGCAGAGTGAATrGrGrG
+
+4. TSO_PCR
+
+Source sequence:
+
+AAGCAGTGGTATCAACGCAGAGT
+
+5. P5-TSO_Hybrid
+
+Source sequence:
+
+AATGATACGGCGACCACCGAGATCTACACGCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+
+6. Nextera_N701
+
+Source sequence:
+
+CAAGCAGAAGACGGCATACGAGATTCGCCTTAGTCTCGTGGGCTCGG
+
+7. Nextera_N702
+
+Source sequence:
+
+CAAGCAGAAGACGGCATACGAGATCTAGTACGGTCTCGTGGGCTCGG
+
+8. Nextera_N703
+
+Source sequence:
+
+CAAGCAGAAGACGGCATACGAGATTTCTGCCTGTCTCGTGGGCTCGG
+
+9. Read1CustomSeqA
+
+Source sequence:
+
+GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTACGT
+
+10. Read1CustomSeqB
+
+Source sequence:
+
+GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTAC
+
+11. P7-TSO_Hybrid
+
+Source sequence:
+
+CAAGCAGAAGACGGCATACGAGATCGTGATCGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+
+12. TruSeq_F
+
+Source sequence:
+
+AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
+
+13. CustSynRNASeq
+
+Source sequence:
+
+CGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGTAC
+
+14. UMI_SMARTdT
+
+Source sequence:
+
+AAGCAGTGGTATCAACGCAGAGTACNNNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTT
+
+Step-by-step library generation
+
+Step 1. Barcoded bead synthesis
+
+Input substrate:
+
+* Toyopearl beads / functionalized microparticles
+
+Added oligos/reagents:
+
+* reverse-direction phosphoramidite synthesis reagents
+
+Molecular event:
+Barcoded primers are synthesized directly on beads. Each bead receives many copies of the same cell barcode through split-and-pool synthesis. After barcode synthesis, each primer receives an 8-base UMI and an oligo-dT capture region.
+
+Product structure:
+
+Bead linker
+
+* PCR handle
+* cell barcode
+* UMI
+* oligo-dT capture sequence
+
+Step 2. Droplet generation and mRNA capture
+
+Input substrate:
+
+* single-cell suspension
+* barcoded primer beads
+
+Added oligos/reagents:
+
+* Drop-seq lysis buffer
+* droplet generation oil
+
+Molecular event:
+Cells and barcoded beads are co-encapsulated in droplets. Cells lyse inside droplets, and polyadenylated mRNAs hybridize to the oligo-dT region of the bead-bound primers.
+
+Product structure:
+
+bead-bound primer
+
+* captured polyadenylated mRNA
+
+Step 3. Droplet breakage and reverse transcription
+
+Input substrate:
+
+* mRNA-bound beads from droplets
+
+Added oligos/reagents:
+
+* Template_Switch_Oligo
+* reverse-transcription reagents
+
+Molecular event:
+Droplets are broken and beads are collected. Reverse transcription is performed on bead-bound mRNAs. Template switching introduces a template-switch sequence downstream of the synthesized cDNA.
+
+Product structure:
+
+PCR handle
+
+* cell barcode
+* UMI
+* oligo-dT
+* cDNA insert
+* template-switch-derived sequence
+
+Step 4. Exonuclease I treatment and PCR amplification
+
+Input substrate:
+
+* reverse-transcribed STAMPs
+
+Added oligos/reagents:
+
+* Exonuclease I
+* TSO_PCR primer
+* KAPA HiFi HotStart ReadyMix
+
+Molecular event:
+Exonuclease I removes unused primers. Bead aliquots are PCR amplified using the template-switch PCR primer.
+
+Product structure:
+
+amplified Drop-seq cDNA
+
+Step 5. Nextera XT tagmentation and library PCR
+
+Input substrate:
+
+* amplified Drop-seq cDNA
+
+Added oligos/reagents:
+
+* Nextera XT reagents
+* P5_TSO_Hybrid
+* Nextera_N701 or other Nextera index primer
+
+Molecular event:
+3' cDNA fragments are prepared for sequencing using Nextera XT tagmentation. Custom primers replace the kit oligos to enrich Drop-seq cDNA fragments and add Illumina sequencing adapters and sample index sequence.
+
+Product structure:
+
+P5 side
+
+* Drop-seq read 1 priming region
+* cell barcode
+* UMI
+* oligo-dT / cDNA insert
+* Nextera read 2 side
+* sample index
+* P7 side
+
+Step 6. Sequencing
+
+Input substrate:
+
+* final Drop-seq sequencing library
+
+Added oligos/reagents:
+
+* Read1CustomSeqA or Read1CustomSeqB, depending on bead sequence version
+
+Molecular event:
+The library is sequenced paired-end. Read 1 is primed with a custom Drop-seq Read 1 primer and reads the cell barcode and UMI. Read 2 reads cDNA sequence for transcript/gene identification.
+
+Read structure:
+
+Read 1: 20 bp total
+
+* bases 1–12: cell barcode
+* bases 13–20: UMI
+
+Read 2:
+
+* cDNA insert sequence
+* 50 bp in the human-mouse experiment
+* 60 bp in the retina experiment
+
+Final canonical library structure
+
+Simplified segment-level structure:
+
+P5 + Drop-seq Read 1 priming region + cell barcode + UMI + oligo-dT / cDNA insert + Nextera Read 2 side + i7 sample index + P7
+
+Sequencing read interpretation:
+
+* Read 1: 12-bp cell barcode + 8-bp UMI
+* Read 2: cDNA insert used for gene identification
+* i7 index read: sample index from Nextera index primer
+
+Human-curation notes
+
+1. The Drop-seq paper describes the bead primer architecture as a PCR handle, cell barcode, UMI, and oligo-dT capture sequence.
+2. The supplemental protocol explicitly provides the oligo sequences in Table S6.
+3. Barcoded Bead SeqA and Barcoded Bead SeqB differ in the sequence immediately upstream of the cell barcode. SeqA is used with Read1CustomSeqA, and SeqB is used with Read1CustomSeqB.
+4. Table S6 shows both Barcoded Bead SeqA and SeqB with a 12-base cell barcode region followed by an 8-base UMI region. If an OCR output shows fewer J characters for SeqB, that should be treated as a text extraction issue.
+5. The source states that Read 1 is 20 bp, with bases 1–12 corresponding to the cell barcode and bases 13–20 corresponding to the UMI.
+6. The protocol uses Nextera XT library preparation with custom primers P5_TSO_Hybrid and Nextera_N701 in place of the standard kit oligos for the main library construction.
+7. Nextera_N702 and Nextera_N703 are printed in Table S6 and used in specific multiplexed/contamination experiments, but Nextera_N701 is the main primer described for standard Drop-seq library construction.
+8. Modified bases such as rG in Template_Switch_Oligo and * in P5_TSO_Hybrid, P7_TSO_Hybrid, and TruSeq_F should be preserved as printed.

drop_seq/drop-seq_paper.pymupdf_text.txt

@@ -0,0 +1,806 @@
+# pymupdf text extraction
+source_pdf: drop-seq_paper.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/drop_seq/drop-seq_paper.pdf
+extraction: PyMuPDF: page.get_text("text", sort=True)
+
+## Page 1
+
+                                        Resource
+
+
+Highly Parallel Genome-wide Expression Profiling of
+Individual Cells Using Nanoliter Droplets
+
+Graphical Abstract                                Authors
+
+                                                     Evan Z. Macosko, Anindita Basu, ...,
+                                                              Aviv Regev, Steven A. McCarroll
+
+                                              Correspondence
+                                                         emacosko@genetics.med.harvard.edu
+                                                                       (E.Z.M.),
+                                                            mccarroll@genetics.med.harvard.edu
+                                                                   (S.A.M.)
+
+                                                           In Brief
+
+                                                            Capturing single cells along with sets of
+                                                              uniquely barcoded primer beads together
+                                                                           in tiny droplets enables large-scale,
+                                                                  highly parallel single-cell transcriptomics.
+                                                            Applying this analysis to cells in mouse
+                                                                           retinal tissue revealed transcriptionally
+                                                                       distinct cell populations along with
+                                                            molecular markers of each type.
+
+
+
+
+
+Highlights                                       Accession Numbers
+
+d Drop-seq enables highly parallel analysis of individual cells   GSE63473
+  by RNA-seq
+
+d Drop-seq encapsulates cells in nanoliter droplets together
+   with DNA-barcoded beads
+
+d Systematic evaluation of Drop-seq library quality using
+   species mixing experiments
+
+d Drop-seq analysis of 44,808 cells identifies 39 cell
+   populations in the retina
+
+
+
+
+
+          Macosko et al., 2015, Cell 161, 1202–1214
+         May 21, 2015 ª2015 Elsevier Inc.
+             http://dx.doi.org/10.1016/j.cell.2015.05.002
+
+## Page 2
+
+                             Resource
+
+
+Highly Parallel Genome-wide Expression Profiling
+of Individual Cells Using Nanoliter Droplets
+
+Evan Z. Macosko,1,2,3,* Anindita Basu,4,5 Rahul Satija,4,6,7 James Nemesh,1,2,3 Karthik Shekhar,4 Melissa Goldman,1,2
+Itay Tirosh,4 Allison R. Bialas,8 Nolan Kamitaki,1,2,3 Emily M. Martersteck,9 John J. Trombetta,4 David A. Weitz,5,10
+Joshua R. Sanes,9 Alex K. Shalek,4,11,12 Aviv Regev,4,13,14 and Steven A. McCarroll1,2,3,*
+1Department of Genetics, Harvard Medical School, Boston, MA 02115, USA
+2Stanley Center for Psychiatric Research, Broad Institute of Harvard and MIT, Cambridge, MA 02142, USA
+3Program in Medical and Population Genetics, Broad Institute of Harvard and MIT, Cambridge, MA 02142, USA
+4Klarman Cell Observatory, Broad Institute of Harvard and MIT, Cambridge, MA 02142, USA
+5School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA
+6New York Genome Center, New York, NY 10013, USA
+7Department of Biology, New York University, New York, NY 10003, USA
+8The Program in Cellular and Molecular Medicine, Children’s Hospital Boston, Boston, MA 02115, USA
+9Department of Molecular and Cellular Biology and Center for Brain Science, Harvard University, Cambridge, MA 02138, USA
+10Department of Physics, Harvard University, Cambridge, MA 02138, USA
+11Ragon Institute of MGH, MIT, and Harvard, Cambridge, MA 02139, USA
+12Institute for Medical Engineering and Science and Department of Chemistry, MIT, Cambridge, MA 02139, USA
+13Department of Biology, MIT, Cambridge, MA 02139, USA
+14Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA
+*Correspondence: emacosko@genetics.med.harvard.edu (E.Z.M.), mccarroll@genetics.med.harvard.edu (S.A.M.)
+ http://dx.doi.org/10.1016/j.cell.2015.05.002
+
+
+
+SUMMARY                                                 work, it will be important to learn the functional capacities and re-
+                                                       sponses of each cell type.
+Cells, the basic units of biological structure and    A major determinant of each cell’s function is its transcriptional
+function, vary broadly in type and state. Single-   program. Recent advances now enable mRNA-seq analysis of
+cell genomics can characterize cell identity and   individual cells (Tang et al., 2009). However, methods of prepar-
+function, but limitations of ease and scale have pre-   ing cells for profiling have been applicable in practice to just hun-
+vented  its broad application. Here we describe   dreds (Hashimshony et al., 2012; Picelli et al., 2013) or (with auto-
+                                                               mation) a few thousand cells (Jaitin et al., 2014), typically after
+Drop-seq, a strategy for quickly profiling thousands
+                                                                       first separating the cells by flow sorting (Shalek et al., 2013) or
+of individual cells by separating them into nanoli-
+                                                               microfluidics (Shalek et al., 2014) and then amplifying each cell’s
+ter-sized aqueous droplets, associating a different                                                              transcriptome separately. Fast, scalable approaches are needed
+barcode with each cell’s RNAs, and sequencing   to characterize complex tissues with many cell types and states,
+them all together. Drop-seq analyzes mRNA tran-   under diverse conditions and perturbations.
+scripts from thousands of individual cells simul-     Here, we describe Drop-seq, a method to analyze mRNA
+taneously while remembering transcripts’  cell of   expression in thousands of individual cells by encapsulating
+origin. We analyzed transcriptomes from 44,808    cells in tiny droplets for parallel analysis. Droplets—nanoliter-
+mouse retinal cells and identified 39 transcription-   scale aqueous compartments formed by precisely combining
+ally  distinct  cell populations, creating a molec-   aqueous and oil flows in a microfluidic device (Thorsen et al.,
+ular atlas of gene expression for known  retinal   2001; Umbanhowar et al., 2000)—have been used as tiny reac-
+                                                                        tion chambers for PCR (Hindson et al., 2011; Vogelstein and
+cell classes and novel candidate  cell subtypes.
+                                                                        Kinzler, 1999) and reverse transcription (Beer et al., 2008). We
+Drop-seq  will accelerate biological discovery by
+                                                          sought here to use droplets to compartmentalize cells into nano-
+enabling routine transcriptional profiling at single-                                                                            liter-sized reaction chambers for analysis of all of their RNAs. A
+cell resolution.                                             basic challenge of using droplets for transcriptomics is to retain
+                                                       a molecular memory of the identity of the cell from which each
+INTRODUCTION                              mRNA transcript was isolated. To accomplish this, we developed
+                                                       a molecular barcoding strategy to remember the cell-of-origin of
+Individual cells are the building blocks of tissues, organs, and or-   each mRNA. We critically evaluated Drop-seq, then used it to
+ganisms. Each tissue contains cells of many types, and cells of   profile cell states along the cell cycle. We then applied it to a com-
+each type can switch among biological states. In most biological   plex neural tissue, mouse retina, and from 44,808 cell profiles
+systems, our knowledge of cellular diversity is incomplete; for   identified 39 distinct populations, each corresponding to one or
+example, the cell-type complexity of the brain is unknown and   a group of closely related cell types. Our results demonstrate
+widely debated (Luo et al., 2008; Petilla Interneuron Nomencla-  how large-scale single-cell analysis can help deepen our under-
+ture Group, et al., 2008). To understand how complex tissues   standing of the biology of complex tissues and cell populations.
+
+
+
+1202  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 3
+
+      A    Complex tissue                   Cell isolation                Cell suspension            STAMPs                      Library
+
+
+
+
+
+                                         Use Drop-Seq to analyze the     Suspend in droplets with         Single-cell transcriptomes     RNA-seq library with 10,000
+                                   RNA of each individual cell        beads (microparticles)         attached to microparticles          single-cell transcriptomes
+
+      B  Barcoded primer bead    C  Synthesis of cell barcode (12 bases)                D  Synthesis of UMI (8 bases)
+                                                                   Synthesis                Synthesis                   Synthesis
+                                                                                                       C                                                       Round 1              Round 2                Round 12                                                                                                                           T    8                                                                                                                                                    rounds
+                                                                                                 + A                                                                                                                                                                                   of                                                                                                 G                                 A            A                                                                     synthesis
+                              G           G
+                                                                                                                                                           Millions of the same cell
+                                 C            C                                         barcode per bead
+                               TTT(T27)          T              T
+            PCR     Cell     UMI                                                                                                 48 different molecular
+                handle  barcode                       0                     4                      16          16,777,216          barcodes (UMIs) per bead
+                                                          Number of unique barcodes in pool
+
+
+Figure 1. Molecular Barcoding of Cellular Transcriptomes in Droplets
+(A) Drop-Seq barcoding schematic. A complex tissue is dissociated into individual cells, which are then encapsulated in droplets together with microparticles
+(gray circles) that deliver barcoded primers. Each cell is lysed within a droplet; its mRNAs bind to the primers on its companion microparticle. The mRNAs are
+reverse-transcribed into cDNAs, generating a set of beads called ‘‘single-cell transcriptomes attached to microparticles’’ (STAMPs). The barcoded STAMPs can
+then be amplified in pools for high-throughput mRNA-seq to analyze any desired number of individual cells.
+(B) Sequence of primers on the microparticle. The primers on all beads contain a common sequence (‘‘PCR handle’’) to enable PCR amplification after STAMP
+formation. Each microparticle contains more than 108 individual primers that share the same ‘‘cell barcode’’ (C) but have different unique molecular identifiers
+(UMIs), enabling mRNA transcripts to be digitally counted (D). A 30-bp oligo dT sequence is present at the end of all primer sequences for capture of mRNAs.
+(C) Split-and-pool synthesis of the cell barcode. To generate the cell barcode, the pool of microparticles is repeatedly split into four equally sized oligonucleotide
+synthesis reactions, to which one of the four DNA bases is added, and then pooled together after each cycle, in a total of 12 split-pool cycles. The barcode
+synthesized on any individual bead reflects that bead’s unique path through the series of synthesis reactions. The result is a pool of microparticles, each
+possessing one of 412 (16,777,216) possible sequences on its entire complement of primers (see also Figure S1).
+(D) Synthesis of a unique molecular identifier (UMI). Following the completion of the ‘‘split-and-pool’’ synthesis cycles, all microparticles are together subjected to
+eight rounds of degenerate synthesis with all four DNA bases available during each cycle, such that each individual primer receives one of 48 (65,536) possible
+sequences (UMIs).
+
+
+
+RESULTS                                             To efficiently generate massive numbers of beads, each with a
+                                                                         distinct barcode, we developed a ‘‘split-and-pool’’ DNA synthe-
+Drop-seq consists of the following steps (Figure 1A): (1) prepare   sis strategy (Figure 1C). A pool of millions of microparticles is
+a single-cell suspension from a tissue; (2) co-encapsulate each   divided into four equally sized groups; a different DNA base
+cell with a distinctly barcoded microparticle (bead) in a nanoli-    (A, G, C, or T) is then added to each. All microparticles are
+ter-scale droplet; (3) lyse cells after they have been isolated in   then re-pooled, mixed, and re-split at random into another four
+droplets; (4) capture a cell’s mRNAs on its companion micropar-   groups, and then a different DNA base (A, G, C, or T) is added
+ticle, forming STAMPs (single-cell transcriptomes attached to   to each of the four new groups. After 12 cycles of split-and-
+microparticles); (5) reverse-transcribe, amplify, and sequence   pool DNA synthesis, the primers on any given microparticle
+thousands of STAMPs in one reaction; and (6) use the STAMP   possess the same one of 412 = 16,777,216 possible 12-bp barc-
+barcodes to infer each transcript’s cell of origin.                 odes, but different microparticles have different sequences
+                                                                    (Figure 1C). The entire microparticle pool then undergoes eight
+A Split-Pool Synthesis Approach to Generate Large        rounds of degenerate oligonucleotide synthesis to generate
+Numbers of Distinctly Barcoded Beads                      the UMI on each oligo (Figure 1D); finally, an oligo-dT sequence
+To deliver large numbers of distinctly barcoded primer mole-   (T30) is synthesized on the 30 end of all oligos on all beads.
+cules into individual droplets, we use microparticles (beads).    To confirm that we could distinguish RNAs based on attached
+We synthesized  oligonucleotide primers  directly on beads   barcodes, we reverse-transcribed a pool of synthetic RNAs onto
+(from 50 to 30, yielding free 30 ends available for enzymatic prim-   11 microparticles and sequenced the resulting cDNAs (Fig-
+ing). Each oligonucleotide is composed of four parts (Figure 1B):   ure S1A and Supplemental Experimental Procedures); 11 micro-
+(1) a constant sequence (identical on all primers and beads) for   particle barcodes each constituted 3.5%–14% of the resulting
+use as a priming site for downstream PCR and sequencing; (2)   sequencing reads, whereas the next-most-abundant 12-mer
+a ‘‘cell barcode’’ (identical across all the primers on the surface   constituted only 0.06% (Figure S1A). These results suggested
+of any one bead, but different from the cell barcodes on other   that the microparticle-of-origin for most cDNAs can be recog-
+beads);  (3) a Unique Molecular Identifier (UMI) (different on   nized by sequencing. We also found that each bead contained
+each primer, to identify PCR duplicates) (Kivioja et al., 2012);  more than 108 barcoded primer sites and that the sequence
+and  (4) an oligo-dT sequence for capturing polyadenylated   complexity of the barcodes approached theoretical limits (Fig-
+mRNAs and priming reverse transcription.                       ures S1B and S1C, Supplemental Experimental Procedures).
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1203
+
+## Page 4
+
+Microfluidics Device for Co-encapsulating Cells with       ranging from 0.36% to 11.3% for the various cell concentrations
+Beads                                                         tested (under the assumption that human-mouse doublets ac-
+We designed a microfluidic ‘‘co-flow’’ device (Utada et al., 2007)   count for half of all doublets). This reflects the greater chance
+to co-encapsulate cells with barcoded microparticles (Figures   at higher cell concentrations that a droplet could encapsulate
+2A and S2 and Data S1). This device quickly co-flows two   multiple  cells. By comparison, previous studies  that used
+aqueous solutions across an  oil channel to form more than  FACS (Jaitin et al., 2014) or a commercial microfluidics platform
+100,000 nanoliter-sized droplets per minute. One flow contains   (Shalek et al., 2014) to isolate single cells reported doublet rates
+the barcoded microparticles suspended in a lysis buffer; the   of 2.3% and 11% respectively, based upon examining micro-
+other flow contains a cell suspension (Figure 2A, left, and 2B).   scopy images of captured cells. In analyzing the above mouse-
+The number of droplets created greatly exceeds the number of  human cell suspension mixture in a commercial microfluidics
+beads or cells injected, so that a droplet will generally contain   system (Fluidigm C1), we found that 30% of the resulting libraries
+zero or one cells, and zero or one beads. Millions of nanoliter-    in that experiment were species-mixed (Figure S3C); about one-
+sized droplets are generated per hour, of which thousands    third of these doublets were visible in the microscopy images.
+contain both a bead and a cell (Movie S1). STAMPs are produced   Single-Cell Impurity
+in the subset of droplets that contain both a bead and a cell.     Species-mixing experiments enabled us to measure single-cell
+                                                                       purity across thousands of libraries prepared at different cell
+Sequencing and Analysis of Many STAMPs in a Single       concentrations. We found that purity was strongly related to
+Reaction                                                                  cell concentration, ranging from 98.8% at 12.5 cells  / ml to
+To efficiently process thousands of STAMPs at once, we break  90.4% at 100 cells / ml (Figure S3B). The largest source of sin-
+droplets, collect the mRNA-bound microparticles, and reverse-    gle-cell impurity appeared to be ambient RNA that is present in
+transcribe the mRNAs (from the microparticle-attached primers)   the cell suspension (a first step of almost all single-cell methods)
+together in one reaction, forming covalent, stable STAMPs (Fig-   and presumably results from cells that are damaged during prep-
+ure 2A, step 7, and Experimental Procedures). A scientist can   aration (Figure S3D). We measured a mean single-cell purity
+then select any desired number of STAMPs for the preparation   of 95.8% for the same cell mixtures in the Fluidigm C1 system
+of 30-end digital expression libraries (Figure 2C, Experimental   (Figure S3C), similar to Drop-seq at 50 cells /ml.
+Procedures). We sequence the resulting molecules from each   Conversion Efficiency
+end (Figure 2C) using high-capacity parallel sequencing. We   The use of synthetic RNA ‘‘spike-in’’ controls at known concen-
+digitally count the number of mRNA transcripts of each gene as-    trations, together with UMIs to avoid double-counting, allows
+certained in each cell, using the UMIs to avoid double-counting   estimation of capture rates for digital single-cell expression tech-
+sequence reads that arose from the same mRNA transcript. We   nologies (Brennecke et al., 2013; Islam et al., 2014). We identified
+thereby create a matrix of digital gene-expression measure-   evidence that PCR and sequencing errors inflate the numbers of
+ments (one measurement per gene per cell) for further analysis   apparently unique UMIs (Table S1 and Supplemental Experi-
+(Figure 2D, Experimental Procedures).                         mental Procedures), so we developed a more conservative esti-
+                                                          mation method than has been used in earlier studies (Islam et al.,
+The Single-Cell Accuracy and Sensitivity of Drop-Seq       2014); in our approach, we collapse similar UMI sequences into a
+Libraries                                                          single count. Using this approach we calculated a capture rate of
+To measure the accuracy with which Drop-seq remembers the  12.8% for Drop-seq (Figure 3G). We corroborated this estimate
+cell-of-origin of each mRNA, we analyzed mixtures of cultured   by making independent digital expression measurements (on
+human (HEK) and mouse (3T3) cells, scoring the numbers of hu-   bulk RNA from 50,000 HEK cells) on ten genes using droplet dig-
+man and mouse transcripts that associated with each cell bar-    ital PCR (ddPCR) (Hindson et al., 2011), calculating an average
+code (Figures 3A, 3B, and S3A). We found that the individual   conversion efficiency of 10.7% (Figures S4A, S4B, and S4C).
+STAMPs created by Drop-seq were highly organism-specific    To further evaluate how the digital transcriptomes ascertained
+(Figures 3A and 3B), indicating high single-cell integrity of the li-   by Drop-seq related to the underlying mRNA content of cells,
+braries. At saturating levels of sequence coverage, we detected  we compared Drop-seq log-expression measurements to those
+an average of 44,295 mRNA transcripts from 6,722 genes in HEK  made by a commonly used in-solution amplification process,
+cells and 26,044 transcripts from 5,663 genes in 3T3 cells (Fig-   finding strong correlation  (r = 0.94, Figure 3E), though Drop-
+ures 3C and 3D).                                         seq ascertained GC-rich transcripts at a lower rate (Figure S4D).
+  To understand how Drop-seq  libraries compare to other  We also compared Drop-seq single-cell log-expression mea-
+single-cell methods, we used three quality metrics: (1) the fre-   surements with measurements from bulk mRNA-seq, observing
+quency of cell-cell doublets; (2) single-cell purity; and (3) tran-   a correlation of r = 0.90 (Figures 3F, S4E, and S4F).
+script capture rates.
+Cell Doublets                                               Cell States: Drop-Seq Analysis of the Cell Cycle
+One potential mode of failure in any single-cell method involves  To evaluate the visibility of cell states in Drop-seq, we first exam-
+cells that stick together or happen to otherwise be co-isolated for   ined cell-to-cell variation among the 589 HEK and 412 3T3
+library preparation. In Drop-seq, across four conditions spanning  STAMPs shown in Figure 3B. Both cultures consisted of asyn-
+12.5  cells/ml to 100  cells/ml, the  fraction of species-mixed   chronously dividing cells; principal components analysis (PCA)
+STAMPs correlated with cell concentration (Figures 3A, 3B,   of the single-cell expression profiles showed the top principal
+and S3B; Experimental Procedures), with cell doublet estimates   components to be dominated by genes with roles in protein
+
+
+
+1204  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 5
+
+      A
+
+
+
+
+
+      B                             C
+
+
+
+
+
+      D
+
+
+
+
+
+Figure 2. Extraction and Processing of Single-Cell Transcriptomes by Drop-Seq
+(A) Schematic of single-cell mRNA-seq library preparation with Drop-seq. A custom-designed microfluidic device joins two aqueous flows before their
+compartmentalization into discrete droplets. One flow contains cells, and the other flow contains barcoded primer beads suspended in a lysis buffer. Immediately
+following droplet formation, the cell is lysed and releases its mRNAs, which then hybridize to the primers on the microparticle surface. The droplets are broken by
+adding a reagent to destabilize the oil-water interface (Experimental Procedures), and the microparticles collected and washed. The mRNAs are then reverse-
+transcribed in bulk, forming STAMPs, and template switching is used to introduce a PCR handle downstream of the synthesized cDNA (Zhu et al., 2001).
+(B) Microfluidic device used in Drop-seq. Beads (brown in image), suspended in a lysis agent, enter the device from the central channel; cells enter from the top
+and bottom. Laminar flow prevents mixing of the two aqueous inputs prior to droplet formation (see also Movie S1). Schematics of the device design and how it is
+operated can be found in Figure S2.
+(C) Molecular elements of a Drop-seq sequencing library. The first read yields the cell barcode and UMI. The second, paired read interrogates sequence from the
+cDNA (50 bp is typically sequenced); this sequence is then aligned to the genome to determine a transcript’s gene of origin.
+(D) In silico reconstruction of thousands of single-cell transcriptomes. Millions of paired-end reads are generated from a Drop-seq library on a high-throughput
+sequencer. The reads are first aligned to a reference genome to identify the gene-of-origin of the cDNA. Next, reads are organized by their cell barcodes,
+and individual UMIs are counted for each gene in each cell (Supplemental Experimental Procedures). The result, shown at far right, is a ‘‘digital expression matrix’’
+in which each column corresponds to a cell, each row corresponds to a gene, and each entry is the integer number of transcripts detected from that gene, in
+that cell.
+
+
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1205
+
+## Page 6
+
+Figure 3. Critical Evaluation of Drop-Seq Using Species-Mixing Experiments
+(A and B) Drop-seq analysis of mixutres of mouse and human cells. Mixtures of human (HEK) and mouse (3T3) cells were analyzed by Drop-seq at the con-
+centrations shown. The scatter plot shows the number of human and mouse transcripts associating to each STAMP. Blue dots indicate STAMPs that were
+designated from these data as human-specifiic (average of 99% human transcripts); red dots indicate STAMPs that were mouse-specific (average 99%). At the
+lower cell concentration, one STAMP barcode (of 570) associated with a mixture of human and mouse transcripts (A, purple). At the higher cell concentration,
+about 1.9% of STAMP barcodes associated with mouse-human mixtures (B). Data for other cell concentrations and a different single-cell analysis platform are in
+Figures S3B and S3C.
+(C and D) Sensitivity analysis of Drop-seq at high read-depth. Violin plots show the distribution of the number of transcripts (C, scored by UMIs) and genes (D)
+detected per cell for 54 HEK (human) STAMPs (blue) and 28 3T3 (mouse) STAMPs (green) that were sequenced to a mean read depth of 737,240 high-quality
+aligned reads per cell.
+(E and F) Correlation between gene expression measurements in Drop-seq and non-single-cell RNA-seq methods. Comparison of Drop-seq gene expression
+measurements (averaged across 550 STAMPs) to measurements from bulk RNA analyzed by: (E) an in-solution template switch amplification (TSA) procedure
+similar to Smart-seq2 (Picelli et al., 2013) (Supplemental Experimental Procedures); and (F) Illumina TruSeq mRNA-seq. All comparisons involve RNA derived from
+the same cell culture flask (3T3 cells). All expression counts were converted to average transcripts per million (ATPM) and plotted as log (1+ATPM).
+(G) Quantitation of Drop-seq capture efficiency by ERCC spike-ins. Drop-seq was performed with ERCC control synthetic RNA at an estimated concentration of
+100,000 ERCC RNA molecules per droplet. 84 beads were sequenced at a mean depth of 2.4 million reads, aligned to the ERCC reference sequences, and UMIs
+counted for each ERCC species, after applying a stringent down-correction for potential sequencing errors (Table S1 and Supplemental Experimental Pro-
+cedures). For each ERCC RNA species above an average concentration of one molecule per droplet, the predicted number of molecules per droplet was plotted
+in log space (x-axis), versus the actual number of molecules detected per droplet by Drop-seq, also in log space (y-axis). Error bars indicate SD. The intercept of a
+regression line, constrained to have a slope of 1 and fitted to the seven highest points, was used to estimate a conversion factor (0.128). A second estimation,
+using the average number of detected transcripts divided by the number of ERCC molecules used (100,000), yielded a conversion factor of 0.125.
+
+
+
+synthesis, growth, DNA replication, and other aspects of the cell   field et al., 2002). We identified 544 human and 668 mouse genes
+cycle. We inferred the cell-cycle phase of each of the 1,001 cells   with expression patterns that varied along the cell cycle (at a
+by scoring for gene sets (signatures) reflecting five phases of the   false discovery rate of 5%; Experimental Procedures) (Figure 4B),
+cell cycle previously characterized in chemically synchronized   including 200 orthologous gene pairs (p < 10 65 by hyper-
+cells (G1/S, S, G2/M, M, and M/G1) (Figure 4A, Table S2) (Whit-   geometric test). Of these orthologous gene pairs, most (82.5%)
+
+
+
+1206  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 7
+
+A                                                                                                                                                           Phase-specific    Figure 4. Cell-Cycle Analysis of HEK and
+   G1/S                                                         G1/S                                                                   score2         3T3 Cells Analyzed by Drop-Seq
+     S                                                   S                                                               1              (A) Cell-cycle state of 589 HEK cells (left) and 412
+                                                                                      3T3 cells (right) measured by Drop-seq. Cells were  G2/M                                                     G2/M                                                               0
+                                                                                                                                          –1          assessed for their progression through the cell
+    M                                         M                                                              –2            cycle by comparison of each cell’s global pattern
+  M/G1                                                     M/G1                                                          of gene expression with gene sets known to
+B   1                                                                                                                                           Avg. normalized   be enriched in one of five phases of the cycle
+                                                                                                                                                 expression
+      2                                                                    1                                                                                                                                           1             (horizontal rows). A phase-specific score was      3
+                                                                    2                                                      calculated for each cell across these five phases
+      4                                                              3                                                               0           (Supplemental Experimental Procedures), and the
+                                                                                                                       cells ordered by their phase scores.
+                                                                    4                                                              –1
+                                                                                                                        (B)                                                                                                     Discovery                                                                                                                              of                                                                                                                                   cell-cycle                                                                                                                                 regulated genes.                                                                                                                                 Heat      5    cluster                                                                    5                                                                         map                                                                                               showing                                                                                                                    the                                                                                                              average normalized                                                                                                                                      expression
+  Gene  6                                                              6                                                          of 544 human and 668 mouse genes found to be
+                                                                                                       regulated by the cell cycle. Maximal and minimal
+                                                                    7
+      7                                                                                              expression was calculated for each gene across a
+                                                                                                                     sliding window of the ordered cells, and compared
+      8                                                              8                                                       with shuffled cells to obtain a false discovery
+            50       150      250      350      450      550                 50    100    150    200    250    300    350    400               rate (FDR) (Experimental Procedures). The plotted
+               Individual human cells (HEK)                         Individual mouse cells (3T3)                  genes (FDR threshold of 5%) were then clustered
+                                                                                        by k-means analysis to identify sets of genes with
+C                                                                                                                  similar expression patterns. Cluster boundaries
+                                         Novel, conserved                                                  are represented by dashed gray lines.
+        Classic cell cycle genes               cell cycle genes
+                                                                                                                     (C)  Representative  cell-cycle  regulated genes
+       CCNB1     MCM6         ATF4       OTUB1                                               discovered by Drop-seq. Selected genes  that
+       CCNB2     MCM7        ARHGAP11A  PARPBP                                          were found to be cell-cycle regulated in both the
+       MCM2      MCM10        ARPC2       RPL26
+                                                                        HEK and 3T3  cell  sets.  Left: genes  that are
+       MCM3      AURKA        CDCA4      SNHG3
+       MCM4      AURKB         E2F7        SRP9                                               well-known to be cell-cycle regulated. Right: some
+       MCM5                     HISTH1E     TCF19                                           genes identified  in this analysis that were not
+                           MCMBP     WDHD1                                                   previously known to be associated with the cell
+                             NCAPG      ZFHX4                                                  cycle  (Experimental  Procedures). A  complete
+                                NXT1
+                                                                                                                                                          list of cell-cycle regulated genes can be found in
+                                                                                                  Table S2.
+
+
+have been previously annotated as related to the cell cycle in at  Embedding (tSNE) (Amir et al., 2013; van der Maaten and Hinton,
+least one species; among the other 17.5%, we found some that   2008). We projected the remaining 36,145 cells in the data into
+would be expected to show cell-cycle variation (e.g., E2F7 and   the tSNE analysis. We then combined a density clustering
+PARPBP) and many that to our knowledge were not previously   approach with post hoc differential expression analysis to divide
+connected to the cell cycle (Figure 4C and Table S2). Single-   44,808 cells among 39 transcriptionally distinct clusters (Supple-
+cell analysis at this scale enabled characterization of cell-cycle   mental Experimental Procedures) ranging from 50 to 29,400
+gene expression without chemical synchronization and at high    cells in size (Figures 5B and 5C). Finally, we organized the 39
+temporal resolution.                                                       cell populations into larger categories (classes) by building a
+                                                      dendrogram of similarity relationships among the 39 cell popula-
+Cell Types: Drop-Seq Analysis of the Retina                   tions (Figure 5D, left).
+We selected the retina as the first tissue to study with Drop-seq    The cell populations inferred from this analysis were readily
+because decades of work has generated molecular information   matched to the known  retinal  cell types, including  all five
+about many retinal cell types (Masland, 2012; Sanes and Zipur-   neuronal cell classes, based on the specific expression of known
+sky, 2010), allowing us to relate our RNA-seq data to prior clas-   markers for these cell types (Figure 5D, right, and Figure S6A).
+sification. The retina contains five neuronal classes—retinal gan-   Additional  clusters corresponded  to astrocytes (associated
+glion, bipolar, horizontal, photoreceptor, and amacrine—each   with retinal ganglion cell axons exiting the retina), resident micro-
+defined by morphological, physiological, and molecular criteria    glia, endothelial cells (from intra-retinal vasculature), pericytes,
+(Figure 5A). Most of the classes are divisible into discrete   and fibroblasts (Figure 5D). The relative abundances of the
+types—a total currently estimated at about 100—but well under   major cell classes in our data agreed with earlier estimates
+half of these types possess known, distinguishing molecular   from microscopy (Jeon et al., 1998) (Table 1).
+markers.
+ We sequenced 49,300 STAMPs prepared from the retinas of   Replication and Cumulative Power of Drop-Seq Data
+14-day-old mice (STAMPs were collected in seven batches   Replication across experimental sessions enables the construc-
+over 4 days). We performed principal components analysis on   tion of cumulatively powerful datasets—but only if data are repli-
+the 13,155 largest libraries (Figure S5, Table S3), then reduced   cable and comparable. The retinal STAMPs were generated on 4
+the 32 statistically significant PCs (Experimental Procedures)   different days (weeks apart), utilizing different litters and multiple
+to two dimensions using t-Distributed Stochastic Neighbor   runs in several sessions, for a total of seven replicates. One of the
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1207
+
+## Page 8
+
+Figure 5. Ab Initio Reconstruction of Retinal Cell Types from 44,808 Single-Cell Transcription Profiles Prepared by Drop-Seq
+(A) Schematic representation of major cell classes in the retina. Photoreceptors (rods or cones) detect light and pass information to bipolar cells, which in turn
+contact retinal ganglion cells that extend axons into other CNS tissues. Amacrine, bipolar and horizontal cells are retinal interneurons; Mu¨ ller glia act as support
+cells for surrounding neurons.
+(B) Clustering of 44,808 Drop-seq single-cell expression profiles into 39 retinal cell populations. The plot shows a two-dimensional representation (tSNE) of global
+gene expression relationships among 44,808 cells; clusters are colored by cell class, according to Figure 5A.
+(C) Differentially expressed genes across 39 retinal cell populations. In this heat map, rows correspond to individual genes found to be selectively upregulated in
+individual clusters (p < 0.01, Bonferroni corrected); columns are individual cells, ordered by cluster (1–39). Clusters with > 1,000 cells were downsampled to 1,000
+cells to prevent them from dominating the plot.
+(D) Gene expression similarity relationships among 39 inferred cell populations. Average expression across all detected genes was calculated for each of 39 cell
+clusters, and the relative (Euclidean) distances between gene-expression patterns for the 39 clusters are represented by a dendrogram. The branches of the
+dendrogram were annotated by examining the differential expression of known markers for retina cell classes and types. Twelve examples are shown at right,
+using violin plots to represent the distribution of expression within the clusters. Violin plots for additional genes are in Figure S6A.
+
+                                                                                                             (legend continued on next page)
+
+1208  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 9
+
+Table 1. Ascertainment of Cell Types and Frequencies in the      Wa¨ssle, 2004). Another amacrine cell population expresses no
+Mouse Retina by Drop-Seq                                  GABAergic, glycinergic or glutamatergic markers;  its neuro-
+                                                                    transmitter is unidentified (nGnG amacrines) (Kay et al., 2011).                                             Percentage of
+                       Percentage of Retina    Cell Population      We first identified markers that were most universally ex-
+Cell Class               (Jeon et al., 1998) (%)    in Drop-Seq (%)     pressed by amacrines relative to other cell classes (Figure 6A).
+Rod photoreceptors      79.9                   65.6          We then assessed the expression of known glycinergic and
+                                                      GABAergic markers; their mutually exclusive expression is a
+Cone photoreceptors      2.1                     4.2
+                                                           fundamental distinction among amacrines. Of the 21 amacrine
+Muller glia                 2.8                     3.6
+                                                                       clusters, 12 were  identifiable as GABAergic (Gad1 and/or
+Retinal ganglion cells       0.5                     1.0                Gad2-positive) and 5 others were glycinergic (glycine transporter
+Horizontal cells            0.5                     0.6                 Slc6a9-positive) (Figure 6B). An additional cell population was
+Amacrine cells             7.0                     9.9                 identified as excitatory by its expression of a glutamate trans-
+Bipolar cells               7.3                   14.0                  porter, Slc17a8 (Figure 6B). The remaining three clusters (clusters
+Microglia        —                       0.2                  4, 20, and 21) had low levels of GABAergic, glycinergic, and glu-
+                                                              tamatergic markers; these likely include nGnG amacrines.Retinal endothelial cells —                       0.6
+                                                Among the glycinergic and GABAergic clusters, we found
+Astrocytes                                         0.1
+                                                many amacrine types with known markers. The most divergent
+The sizes of the 39 annotated cell clusters produced from Drop-seq were
+                                                                    glycinergic cluster appeared to correspond to the A-II amacrine
+used to estimate their fractions of the total cell population. These data
+                                                          neurons (Figure 6B, cluster 16), as this was the only cluster towere compared with those obtained by microscopy techniques (Jeon
+et al., 1998).                                                        strongly express the Gjd2 gene encoding the gap junction pro-
+                                                                         tein connexin 36 (Feigenspan et al., 2001). Ebf3, a transcription
+                                                                     factor found in SEG glycinergic as well as nGnG amacrines,
+runs was performed at a particularly low cell concentration (15  was specific to clusters 17 and 20. Starburst amacrine neurons
+cells/ml) and thus high purity, to evaluate whether results were ar-   (SACs), the only retinal cells that use acetylcholine as a co-trans-
+tifacts of cell-cell doublets or single-cell impurity. We found that   mitter, were identifiable as cluster 3 by their expression of the
+all 39 clusters contained cells from every experiment. One clus-   cholinergic marker Chat (Figure 6B). Unlike other GABAergic
+ter (arrow in Figure 5E; star in Figure S6B), which drew dispropor-    cells, SACs expressed Gad1 but not Gad2, as previously
+tionately from two replicates, expressed markers of fibroblasts, a   observed in rabbit (Famiglietti and Sundquist, 2010).
+non-retinal cell type that is present in tissue surrounding the   We then identified selectively expressed markers for each of the
+retina, and hence likely represents imprecise dissection.        21 amacrine cell populations (Figure 6C and Table S4). We vali-
+ We examined how the classification of cells (based on their   dated two of the markers immunohistochemically. First, we co-
+patterns of gene expression) evolved as a function of the   stained retinal sections with antibodies to the transcription factor
+numbers of cells in analysis. We used 500, 2,000, or 9,731 cells  MAF, the top marker of cluster 7, plus antibodies to either GAD1 or
+from our dataset, and asked how (for example) cells identified as   SLC6A9, markers of GABAergic and glycinergic transmission,
+amacrines in the full dataset clustered in analyses of smaller   respectively. As predicted by the Drop-seq analysis, MAF was
+numbers of cells (Figure 5F). As the number of cells in the data   found in a small subset of amacrine cells that were GABAergic
+increased, distinctions between related clusters become clearer,   and not glycinergic (Figure 6D). Cluster 7 had numerous genes
+stronger, and finer in resolution, with the result that a greater   that were enriched relative to its nearest neighbor, cluster 6 (Fig-
+number of rare amacrine cell sub-populations (each represent-   ure 6E, 16 genes > 2.8-fold enrichment, p < 10  9), including
+ing 0.1%–0.9% of the cells in the experiment) could ultimately   Crybb3, which belongs to the crystallin family of proteins that
+be distinguished from one another (Figure 5F).                   are known to be directly upregulated by Maf (Yang and Cvekl,
+                                                                  2005), and another, the protease Mmp9, which accepts crystallins
+Profiles of Amacrine Cell Types                           as substrates (Descamps et al., 2005). Second, we stained sec-
+To characterize distinctions among closely related cell popula-   tions with antibodies to PPP1R17 (Figure 6F), a nominated marker
+tions, we focused on the 21 clusters of amacrines. Amacrines   of cluster 20. Cluster 20 shows weak, infrequent glycine trans-
+are the most morphologically diverse neuronal class (Masland,   porter expression and is one of only two clusters (with cluster
+2012), but the majority of types lack defining molecular markers.   21) that express Neurod6, a marker of nGnG neurons (Kay et al.,
+Most amacrine cells are inhibitory,  utilizing either GABA or   2011). We used a transgenic strain (MitoP) that has been shown
+glycine as a neurotransmitter. Excitatory amacrine cells that   to express CFP specifically in nGnG amacrines (Kay et  al.,
+release glutamate have also been identified (Haverkamp and   2011). PPP1R17 stained 85% of all CFP-positive amacrines in
+
+
+
+(E) Representation of experimental replicates in each cell population. tSNE plot from Figure 2B, with each cell now colored by experimental replicate (for visual
+clarity, the central rod cluster was downsampled to 10,000 cells). Each of the seven replicates contributes to all 39 cell populations. Cluster 36 (arrow), in which
+these replicates are unevenly represented, expressed markers of fibroblasts, which are not native to the retina and are presumably a dissection artifact (see also
+Figure S6B).
+(F) Trajectory of amacrine clustering as a function of number of cells analyzed. Three different downsampled datasets were generated: (1) 500, (2) 2,000, or (3)
+9,731 cells (Supplemental Experimental Procedures). Cells identified as amacrines (clusters 3–23) in the full analysis are here colored by their cluster identities in
+that analysis. Analyses of smaller numbers of cells incompletely distinguished these subpopulations from one another.
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1209
+
+## Page 10
+
+Figure 6. Finer-Scale Expression Distinctions among Amacrine Cells, Cones, and Retinal Ganglion Cells
+(A) Pan-amacrine markers. The expression levels of the six genes identified (Nrxn2, Atp1b1, Pax6, Slc32a1, Slc6a1, Elavl3) are represented as dot plots across all
+39 clusters; larger dots indicate broader expression within the cluster; deeper red denotes a higher expression level.
+
+                                                                                                             (legend continued on next page)
+
+
+
+1210  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 11
+
+the MitoP line, validating this as a marker of nGnG cells (Figure 6F).      In validating Drop-seq, we developed stringent species-mix-
+PPP1R17 was one of several markers that distinguished Cluster   ing experiments to measure single-cell purity and cell doublet
+20 from its closest neighbor, Cluster 21 (Figure 6G; 12 genes >   rates in our libraries. In another article in this issue, Klein et al.
+2.8-fold enrichment, p < 10  9). The differences between Clusters   (Klein et al., 2015) describe a droplet-based approach to sin-
+20 and 21 suggest a hitherto unsuspected level of heterogeneity    gle-cell RNA-seq and also use species-mixing experiments to
+among nGnG amacrines.                                        evaluate  it. Our results indicate that all methods of isolating
+                                                                    single cells from a cell suspension, including Drop-seq, fluores-
+Supervised Analysis Reveals Additional Diversity          cence activated cell sorting (FACS) and microfluidics, are vulner-
+Our unsupervised analysis grouped cells into 39 transcription-   able to impurities, and highlight the value of performing species
+ally  distinct  populations, but morphological and  functional   mixing experiments to assess single-cell approaches. In our
+criteria suggest that there are  100 retinal cell types. We asked   retina analysis, even relatively impure libraries generated in
+whether supervised analysis could reveal multiple types within   ‘‘ultra-high-throughput’’ modes (100 cells per ml, allowing the
+individual clusters. For example, retinal ganglion cells (RGCs),   processing of 10,000 cells per hour at 10% doublet and impu-
+which consist of about 30 types (Sanes and Masland, 2015),    rity rates) appeared to yield a robust and biologically validated
+formed a single cluster in our analysis, perhaps because  it is    cell classification, but other tissues or applications may require
+a rare cell population (1%, Table 1). Five RGC types, called   using Drop-seq in purer modes.
+intrinsically photosensitive RGCs (ipRGCs), express Opn4, the     Unsupervised  computational  analysis  of  Drop-seq  data
+gene encoding the photopigment melanopsin. Opn4+ RGCs   identified 39 transcriptionally distinct retinal cell populations,
+(26/432) expressed nine genes at levels 2-fold higher than  many representing specific subtypes of the major retinal cell
+Opn4- RGCs  (p < 109, Figure 6H), including Tbr2/Eomes,   classes (Figures 5 and  6).  It  is a particular strength of the
+known to be a selective marker for this population (Sweeney   retina that establishing correspondence between cluster and
+et al., 2014). This result reveals additional heterogeneity that   type was in many cases straightforward; an important direc-
+may also emerge ab  initio as analyses expand to include   tion will be to identify cell types and states in other parts of
+more cells.                                                     the brain—as well as in other tissues—about which less  is
+                                                                     currently known.
+DISCUSSION                                 We see many applications of Drop-seq, beyond the identifica-
+                                                                        tion of cell types and cell states. Genome-scale genetic studies
+Ascertaining transcriptional variation across individual cells is a   are identifying many genes whose variation contributes to disease
+valuable way of learning about complex tissues and functional    risk, but biology has lacked similarly high-throughput ways of
+responses, but single-cell analysis has been limited by the time   connecting these genes to specific cell populations and unique
+and cost of preparing libraries from many individual cells. A sci-   functional responses. Drop-seq could be used to provide initial in-
+entist employing Drop-seq can prepare 10,000 single-cell  li-   sights into how these genes function in the diverse cell types
+braries for sequencing in 12 hr, for about 6.5 cents per cell (Table   composing each tissue. In addition, coupling Drop-seq to pertur-
+S5), representing a >100-fold improvement in both time and cost   bations—suchassmallmolecules, mutations,pathogens,orother
+relative to existing methods. A Drop-seq setup can be con-   stimuli—could generate an information-rich, multi-dimensional
+structed quickly and inexpensively in a standard biology lab us-   readout of the influence of perturbations on many kinds of cells.
+ing readily available equipment (Figure S2B and Supplemental    The functional implications of a gene’s expression are a prod-
+Experimental Procedures). We hope that ease, speed, and low   uct not just of that gene’s intrinsic properties, but also of the
+cost facilitate exuberant experimentation, careful replication,   entire cell-level context in which the gene is expressed. We
+and many cycles of experiments, analyses, ideas, and more   hope Drop-seq enables the abundant and routine discovery of
+experiments.                                            such relationships in many areas of biology.
+
+
+
+(B) Identification of known amacrine types among clusters. The 21 amacrine clusters consisted of 12 GABAergic, five glycinergic, one glutamatergic, and three
+non-GABAergic non-glycinergic clusters. Starburst amacrines were identified in cluster 3 by their expression of Chat; excitatory amacrines by expression of
+Slc17a8; A-II amacrines by their expression of Gjd2; and SEG amacrine neurons by their expression of Ebf3.
+(C) Nomination of novel candidate markers of amacrine subpopulations. Each cluster was screened for genes differentially expressed in that cluster relative to all
+other amacrine clusters (p < 0.01, Bonferroni corrected) (McDavid et al., 2013), and filtered for those with highest relative enrichment. Expression of a single
+candidate marker for each cluster is shown across all amacrines.
+(D) Validation of MAF as a marker for a GABAergic amacrine population. Staining of a fixed adult retina from wild-type mice for MAF (i, ii, v, and green staining in iv
+and vii), GAD1 (iii and iv, red staining), and SLC6A9 (vi and vii, red staining), demonstrating co-localization of MAF with GAD1, but not SLC6A9.
+(E) Differential expression of cluster 7 (Maf+) with nearest neighboring amacrine cluster (#6). Average gene expression was compared between cells in clusters 6
+and 7; 16 genes (red dots) were identified with >2.8-fold enrichment in cluster 7 (p < 10  9).
+(F) Validation of PPP1R17 as a marker for an amacrine subpopulation. Staining of a fixed adult retina from Mito-P mice, which express CFP in both nGnG
+amacrines and type 1 bipolars (Kay et al., 2011). Overlapping labeling by PPP1R17 antibody (green) and Mito-P CFP (red) supports Drop-seq identification of
+Ppp1r17 expression in the nGnG amacrine neurons. 85% of CFP+ cells were PPP1R17+ and 50% of the PPP1R17+ cells were CFP   , suggesting a second
+amacrine type expressing this marker. Blue staining is for VSX2, a marker of bipolar neurons.
+(G) Differential expression of cluster 20 (Ppp1r17+) with nearest neighboring amacrine cluster (#21). Average gene expression was compared between cells in
+clusters 20 and 21; 12 genes (red dots) were identified with >2.8-fold enrichment in cluster 20 (p < 10  9).
+(H) Differential expression of melanopsin-positive and negative RGCs. Average expression was compared between Opn4-positive and -negative RGCs in cluster
+2. Seven genes were identified as enriched in Opn4-positive cells (red dots, > 2-fold, p < 10  9).
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1211
+
+## Page 12
+
+EXPERIMENTAL PROCEDURES                                              igates the impact of noisy variation in the lower complexity libraries due to
+                                                                 gene dropouts. It was also reliable in the sense that when we withheld from
+Device Design and Fabrication                                            the t-SNE all cells from a given cluster and then tried to project them, these
+Microfluidic devices were designed using AutoCAD software (Autodesk), and    withheld cells were not spuriously assigned to another cluster by the projection
+the components tested using COMSOL Multiphysics (COMSOL). Full details    (Table S7). Point clouds on the t-SNE map represent candidate cell types; den-
+are described in Supplemental Experimental Procedures.                           sity clustering (Ester et al., 1996) identified these regions. Differential expres-
+                                                                              sion testing (McDavid et al., 2013) was then used to confirm that clusters
+Barcoded Microparticle Synthesis                                   were distinct from each other. Hierarchical clustering based on Euclidean dis-
+Bead functionalization and reverse-direction phosphoramidite synthesis were   tance and complete linkage was used to build a tree relating the clusters. We
+performed by Chemgenes Corp (Wilmington, MA). ‘‘Split-and-pool’’ cycles   noted expression of several rod-specific genes, such as Rho and Nrl, in every
+were accomplished by removing the dry resin from each column, hand mixing,     cell cluster, an observation that has been made in another retinal cell gene
+and weighing out four equal portions before returning the resin for an additional    expression study (Siegert et al., 2012) and likely arises from solubilization
+cycle of synthesis. Full details are described in Supplemental Experimental    of these high-abundance transcripts during  cell suspension preparation.
+Procedures.                                                                    Additional information regarding retinal cell data analysis can be found in the
+                                                                      Supplemental Experimental Procedures.
+Drop-Seq Procedure
+Monodisperse droplets  1 nl in size were generated using the microfluidic de-
+                                                     ACCESSION NUMBERS
+vice described in Supplemental Experimental Procedures, in which barcoded
+microparticles, suspended in lysis buffer, were flowed at a rate equal to that of
+                                                                 The accession number for the raw and analyzed data reported in this paper is
+a single-cell suspension, so that resulting droplets were composed of an equal
+                                                            GEO: GSE63473.
+amount of each component. As soon as droplet generation was complete,
+droplets were broken with perfluorooctanol in 30 ml of 63 SSC. The addition
+of a large aqueous volume to the droplets reduces hybridization events after  SUPPLEMENTAL INFORMATION
+droplet breakage, because DNA base pairing follows second-order kinetics
+(Britten and Kohne, 1968; Wetmur and Davidson, 1968). The beads were   Supplemental Information includes Supplemental Experimental Procedures,
+then washed and resuspended in a reverse transcriptase mix, followed by a    six figures, seven tables, one movie, and one data file and can be found with
+treatment with exonuclease  I to remove unextended primers. The beads    this article online at http://dx.doi.org/10.1016/j.cell.2015.05.002.
+were then washed, counted, aliquoted into PCR tubes, and PCR amplified.
+The PCR reactions were purified and pooled, and the amplified cDNA quanti-
+                                                  AUTHOR CONTRIBUTIONS
+fied on a BioAnalyzer High Sensitivity Chip (Agilent). The cDNA was frag-
+mented and amplified for sequencing with the Nextera XT DNA sample prep
+                                                                         E.Z.M. developed the barcoding and molecular biology analysis, advised by
+kit (Illumina) using custom primers that enabled the specific amplification of
+                                                                        S.A.M. A.B. designed and fabricated the microfluidic devices, advised byonly the 30 ends (Table S6). The libraries were purified, quantified, and then
+                                                                     D.A.W. and A.R. E.Z.M. and M.G. developed Drop-seq experimental protocols
+sequenced on the Illumina NextSeq 500.  All  details regarding reaction
+                                                                and performed the Drop-seq experiments in S.A.M.’s lab. J.N. developed the
+conditions, primers used, and sequencing specifications can be found in the
+                                                                methods and software for obtaining digital gene expression measurements for
+Supplemental Experimental Procedures.
+                                                                  each cell, advised by E.Z.M. and S.A.M. J.N., E.Z.M. and S.A.M. performed the
+                                                                          analyses of species-mixing experiments. I.T. performed the cell-cycle anal-
+Cell-Cycle Analysis of HEK and 3T3 Cells
+                                                                                          ysis. A.R.B. prepared the retinal cell suspensions. R.S., K.S., and A.R. devel-
+Gene sets reflecting five phases of the HeLa cell cycle (G1/S, S, G2/M, M and
+                                                               oped and performed the retinal cell type clustering analyses with contribution
+M/G1) were taken from Whitfield et al. (Whitfield et al., 2002) with some modi-
+                                                                      from N.K. E.Z.M., R.S., K.S., and J.R.S. interpreted the retina expression data.
+fication (Supplemental Experimental Procedures and Table S2). A phase-spe-
+                                                                   E.M.M. and J.R.S. performed the immunohistochemistry experiments. J.J.T.
+cific score was generated for each cell, across all five phases, using averaged
+                                                                and A.K.S. performed the Fluidigm C1 experiments. E.Z.M., S.A.M., A.R.,
+normalized expression levels (log2(TPM+1)) of the genes in each set. Cells
+                                                                                   A.B., and A.K.S. conceived the study and key ways that Drop-seq works
+were then ordered along the cell cycle by comparing the patterns of these
+                                                                             together as an integrated system. E.Z.M. and S.A.M. wrote the manuscript
+five phase scores per cell. To identify cell-cycle-regulated genes, we used a
+                                                                              with contributions from all authors.
+sliding window approach, and identified windows of maximal and minimal
+average expression, both for ordered cells, and for shuffled cells, to evaluate
+the false-discovery rate. Full details may be found in Supplemental Experi-  ACKNOWLEDGMENTS
+mental Procedures.
+                                                                              This work was supported by the Stanley Center for Psychiatric Research (to
+Principal Components and Clustering Analysis of Retina Data             S.M.), the MGH Psychiatry Residency Research Program and Stanley-MGH
+The clustering algorithm for the retinal cell data was implemented and per-    Fellowship in Psychiatric Neuroscience (to E.Z.M.), a Stewart Trust Fellows
+formed using Seurat, a recently developed R package for single-cell analysis   Award (to S.M.), a grant from the Simons Foundation to the Simons Center
+(Satija et al., 2015). PCA was first performed on a 13,155-cell ‘‘training set’’    for the Social Brain at MIT (to A.R., S.M., and D.W.), an NHGRI CEGS P50
+of the 49,300-cell dataset, using single-cell libraries in which transcripts from   HG006193 (to A.R.), the Klarman Cell Observatory (to A.R. and A.B.), NIMH
+>900 genes were detected. We found this approach was more effective in    grant U01MH105960 (to S.M., A.R. and J.R.S.), NIMH grant R25MH094612
+discovering structures corresponding to rare cell types than performing PCA     (to E.M.), NIH F32 HD075541 (to R.S.). AR is an investigator of the Howard
+on the full dataset, which was dominated by numerous, tiny rod photorecep-   Hughes Medical Institute. Microfluidic device fabrication was performed at
+tors (Supplemental Experimental Procedures). Thirty-two statistically signifi-    the Harvard Center for Nanoscale Systems (CNS), a member of the National
+cant PCs were  identified using a permutation  test and independently   Nanotechnology Infrastructure Network (National Science Foundation award
+confirmed using a modified resampling procedure (Chung and Storey, 2015).    no. ECS-0335765), with support from the National Science Foundation
+We projected individual cells within the training set based on their PC scores   (DMR-1310266) and the Harvard Materials Research Science and Engineering
+onto a single two-dimensional map using t-Distributed Stochastic Neighbor   Center (DMR-1420570). We thank Christina Usher and Leslie Gaffney for con-
+Embedding (t-SNE) (van der Maaten and Hinton, 2008). The remaining    tributions to the manuscript figures and Chris Patil for helpful comments on the
+36,145 single-cell libraries (<900 genes detected) were next projected on    manuscript. We thank Connie Cepko for helpful conversations about the retina
+this t-SNE map, based on their representation within the PC-subspace of    data, Beth Stevens for advice on retinal dissociations, and Assaf Rotem and
+the training set (Berman et al., 2014; Shekhar et al., 2014). This approach mit-   Huidan Zhang for advice on microfluidics design and fabrication. A.R. is a
+
+
+
+1212  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.
+
+## Page 13
+
+member of the Scientific Advisory Board for Thermo Fisher Scientific and   Kay, J.N., Voinescu, P.E., Chu, M.W., and Sanes, J.R. (2011). Neurod6 expres-
+Syros Pharmaceuticals and a consultant for Driver Genomics.                  sion defines new retinal amacrine cell subtypes and regulates their fate. Nat.
+                                                                            Neurosci. 14, 965–972.
+Received: November 9, 2014                                                         Kivioja, T., Va¨ ha¨ rautio, A., Karlsson, K., Bonke, M., Enge, M., Linnarsson, S.,
+Revised: March 4, 2015                                             and Taipale, J. (2012). Counting absolute numbers of molecules using unique
+Accepted: April 30, 2015                                                   molecular identifiers. Nat. Methods 9, 72–74.
+Published: May 21, 2015
+                                                                                       Klein, A.M., Mazutis, L., Akartuna, I., Tallapragada, N., Veres, A., Li, V., Pesh-
+REFERENCES                                                                          kin, L., Weitz, D.A., and Kirschner, M.W. (2015). Droplet barcoding for single
+                                                                                               cell transcriptomics and its application to embryonic stem cells. Cell 161,
+Amir, A.D., Davis, K.L., Tadmor, M.D., Simonds, E.F., Levine, J.H., Bendall,    this issue, 1187–1201.
+S.C., Shenfeld, D.K., Krishnaswamy, S., Nolan, G.P., and Pe’er, D. (2013).    Luo, L., Callaway, E.M., and Svoboda, K. (2008). Genetic dissection of neural
+viSNE enables visualization of high dimensional single-cell data and reveals    circuits. Neuron 57, 634–660.
+phenotypic heterogeneity of leukemia. Nat. Biotechnol. 31, 545–552.
+                                                                      Masland, R.H. (2012). The neuronal organization of the retina. Neuron 76,
+Beer, N.R., Wheeler, E.K., Lee-Houghton, L., Watkins, N., Nasarabadi, S.,   266–280.
+Hebert, N., Leung, P., Arnold, D.W., Bailey, C.G., and Colston, B.W. (2008).
+                                                                    McDavid, A., Finak, G., Chattopadyay, P.K., Dominguez, M., Lamoreaux, L.,
+On-chip single-copy real-time reverse-transcription PCR in isolated picoliter
+                                                                Ma, S.S., Roederer, M., and Gottardo, R. (2013). Data exploration, quality con-
+droplets. Anal. Chem. 80, 1854–1858.
+                                                                                                    trol and testing in single-cell qPCR-based gene expression experiments. Bio-
+Berman, G.J., Choi, D.M., Bialek, W., and Shaevitz, J.W. (2014). Mapping the                                                                                informatics 29, 461–467.
+stereotyped behaviour of freely moving fruit flies. J. R. Soc. Interface 11,
+                                                                                              Petilla Interneuron Nomenclature Group, Ascoli, G.A., Alonso-Nanclares, L.,20140672.
+                                                                       Anderson, S.A., Barrionuevo, G., Benavides-Piccione, R., Burkhalter, A., Buz-
+Brennecke, P., Anders, S., Kim, J.K., Ko1odziejczyk, A.A., Zhang, X., Proser-
+                                                                                                        sa´ki, G., Cauli, B., Defelipe, J., Faire´ n, A., et al. (2008). Petilla terminology:
+pio, V., Baying, B., Benes, V., Teichmann, S.A., Marioni, J.C., and Heisler,
+                                                                       nomenclature of features of GABAergic interneurons of the cerebral cortex.
+M.G. (2013). Accounting for technical noise in single-cell RNA-seq experi-
+                                                                              Nat. Rev. Neurosci. 9, 557–568.
+ments. Nat. Methods 10, 1093–1095.
+                                                                                                    Picelli, S., Bjo¨ rklund, A.K., Faridani, O.R., Sagasser, S., Winberg, G., and
+Britten, R.J., and Kohne, D.E. (1968). Repeated sequences in DNA. Hundreds
+                                                                      Sandberg, R.  (2013). Smart-seq2  for  sensitive  full-length transcriptome
+of thousands of copies of DNA sequences have been incorporated into the
+                                                                                profiling in single cells. Nat. Methods 10, 1096–1098.
+genomes of higher organisms. Science 161, 529–540.
+Chung, N.C., and Storey, J.D. (2015). Statistical Significance of Variables   Sanes, J.R., and Masland, R.H. (2015). The Types of Retinal Ganglion Cells:
+Driving Systematic Variation in High-Dimensional Data. Bioinformatics 31,    Current Status and Implications for Neuronal Classification. Annu. Rev. Neuro-
+545–554.                                                                                   sci. Published online April 9, 2015.
+Descamps, F.J., Martens, E., Proost, P., Starckx, S., Van den Steen, P.E., Van   Sanes, J.R., and Zipursky, S.L. (2010). Design principles of insect and verte-
+Damme, J., and Opdenakker, G. (2005). Gelatinase B/matrix metalloprotei-    brate visual systems. Neuron 66, 15–36.
+nase-9 provokes cataract by cleaving lens betaB1 crystallin. FASEB J. 19,    Satija, R., Farrell, J.A., Gennert, D., Schier, A.F., and Regev, A. (2015). Spatial
+29–35.                                                                        reconstruction of single-cell gene expression data. Nat. Biotechnol. Published
+Ester, M., Kriegel, H.P., Sander, J., and Xu, X. (1996). A density-based algo-    online 13 April, 2015. http://dx.doi.org/10.1038/nbt.3192.
+rithm for discovering clusters in large spatial databases with noise (Menlo    Shalek,  A.K.,  Satija,  R., Adiconis,  X., Gertner,  R.S., Gaublomme,  J.T.,
+Park, Calif: AAAI Press).                                               Raychowdhury, R., Schwartz, S., Yosef, N., Malboeuf, C., Lu, D., et  al.
+Famiglietti, E.V., and Sundquist, S.J. (2010). Development of excitatory and    (2013).  Single-cell transcriptomics reveals bimodality  in expression and
+inhibitory neurotransmitters in transitory cholinergic neurons, starburst ama-    splicing in immune cells. Nature 498, 236–240.
+crine cells, and GABAergic amacrine cells of rabbit retina, with implications
+                                                                            Shalek, A.K., Satija, R., Shuga, J., Trombetta, J.J., Gennert, D., Lu, D., Chen,
+for previsual and visual development of retinal ganglion cells. Vis. Neurosci.
+                                                                                         P., Gertner, R.S., Gaublomme,  J.T., Yosef, N., et  al. (2014). Single-cell
+27, 19–42.
+                                                              RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510,
+Feigenspan, A., Teubner, B., Willecke, K., and Weiler, R. (2001). Expression   363–369.
+of neuronal connexin36  in  AII amacrine  cells of the mammalian  retina.
+                                                                        Shekhar, K., Brodin, P., Davis, M.M., and Chakraborty, A.K. (2014). Automatic
+J. Neurosci. 21, 230–239.
+                                                                              Classification of Cellular Expression by Nonlinear Stochastic Embedding
+Hashimshony, T., Wagner, F., Sher, N., and Yanai, I. (2012). CEL-Seq: single-                                                                  (ACCENSE). Proc. Natl. Acad. Sci. USA 111, 202–207.
+cell RNA-Seq by multiplexed linear amplification. Cell Rep. 2, 666–673.
+                                                                                    Siegert, S., Cabuy, E., Scherf, B.G., Kohler, H., Panda, S., Le, Y.Z., Fehling,
+Haverkamp, S., and Wa¨ ssle, H. (2004). Characterization of an amacrine cell
+                                                                                      H.J., Gaidatzis, D., Stadler, M.B., and Roska, B. (2012). Transcriptional code
+type of the mammalian retina immunoreactive for vesicular glutamate trans-
+                                                                and disease map for adult retinal cell types. Nat. Neurosci. 15, 487–495,
+porter 3. J. Comp. Neurol. 468, 251–263.
+                                                                    S1–S2.
+Hindson, B.J., Ness, K.D., Masquelier, D.A., Belgrader, P., Heredia, N.J.,
+                                                                  Sweeney, N.T., Tierney, H., and Feldheim, D.A. (2014). Tbr2 is required to
+Makarewicz, A.J., Bright, I.J., Lucero, M.Y., Hiddessen, A.L., Legler, T.C.,
+                                                                         generate a neural circuit mediating the pupillary light reflex. J. Neurosci. 34,
+et al. (2011). High-throughput droplet digital PCR system for absolute quanti-
+                                                                     5447–5453.
+tation of DNA copy number. Anal. Chem. 83, 8604–8610.
+Islam, S., Zeisel, A., Joost, S., La Manno, G., Zajac, P., Kasper, M., Lo¨nner-   Tang, F., Barbacioru, C., Wang, Y., Nordman, E., Lee, C., Xu, N., Wang, X.,
+berg, P., and Linnarsson, S. (2014). Quantitative single-cell RNA-seq with   Bodeau, J., Tuch, B.B., Siddiqui, A., et al. (2009). mRNA-Seq whole-transcrip-
+unique molecular identifiers. Nat. Methods 11, 163–166.                   tome analysis of a single cell. Nat. Methods 6, 377–382.
+Jaitin, D.A., Kenigsberg, E., Keren-Shaul, H., Elefant, N., Paul, F., Zaretsky, I.,    Thorsen, T., Roberts, R.W., Arnold, F.H., and Quake, S.R. (2001). Dynamic
+Mildner, A., Cohen, N., Jung, S., Tanay, A., and Amit, I. (2014). Massively par-    pattern formation in a vesicle-generating microfluidic device. Phys. Rev.
+allel single-cell RNA-seq for marker-free decomposition of tissues into cell    Lett. 86, 4163–4166.
+types. Science 343, 776–779.                                        Umbanhowar, P.B., Prasad, V., and Weitz, D.A. (2000). Monodisperse Emul-
+Jeon, C.J., Strettoi, E., and Masland, R.H. (1998). The major cell populations of    sion Generation via Drop Break Off in a Coflowing Stream. Langmuir 16,
+the mouse retina. J. Neurosci. 18, 8936–8946.                              347–351.
+
+
+
+                                                                                Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.  1213
+
+## Page 14
+
+Utada, A.S., Fernandez-Nieves, A., Stone, H.A., and Weitz, D.A. (2007).    Whitfield, M.L., Sherlock, G., Saldanha, A.J., Murray, J.I., Ball, C.A., Alexander,
+Dripping to jetting transitions in coflowing liquid streams. Phys. Rev. Lett.    K.E., Matese, J.C., Perou, C.M., Hurt, M.M., Brown, P.O., and Botstein, D.
+99, 094502.                                                                       (2002). Identification of genes periodically expressed in the human cell cycle
+                                                                and their expression in tumors. Mol. Biol. Cell 13, 1977–2000.
+van der Maaten, L., and Hinton, G. (2008). Visualizing Data using t-SNE.
+J. Mach. Learn. Res. 9, 2579–2605.                                       Yang, Y., and Cvekl, A. (2005). Tissue-specific regulation of the mouse alphaA-
+                                                                                            crystallin gene in lens via recruitment of Pax6 and c-Maf to its promoter. J. Mol.
+Vogelstein, B., and Kinzler, K.W. (1999). Digital PCR. Proc. Natl. Acad. Sci.    Biol. 351, 453–469.
+USA 96, 9236–9241.
+                                                                        Zhu, Y.Y., Machleder, E.M., Chenchik, A., Li, R., and Siebert, P.D. (2001).
+Wetmur, J.G., and Davidson, N. (1968). Kinetics of renaturation of DNA. J. Mol.   Reverse transcriptase template switching: a SMART approach for full-length
+Biol. 31, 349–370.                                           cDNA library construction. Biotechniques 30, 892–897.
+
+
+
+
+
+1214  Cell 161, 1202–1214, May 21, 2015 ª2015 Elsevier Inc.

drop_seq/drop-seq_supp.pymupdf_text.txt

@@ -0,0 +1,1470 @@
+# pymupdf text extraction
+source_pdf: drop-seq_supp.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/drop_seq/drop-seq_supp.pdf
+extraction: PyMuPDF: page.get_text("text", sort=True)
+
+## Page 1
+
+Cell
+
+Supplemental Information
+Highly Parallel Genome-wide Expression Profiling
+
+of Individual Cells Using Nanoliter Droplets
+
+Evan Z. Macosko, Anindita Basu, Rahul Satija, James Nemesh, Karthik Shekhar,
+
+Melissa Goldman, Itay Tirosh, Allison R. Bialas, Nolan Kamitaki, Emily M. Martersteck,
+
+John J. Trombetta, David A. Weitz, Joshua R. Sanes, Alex K. Shalek, Aviv Regev,
+
+Steven A. McCarroll
+
+## Page 2
+
+Supplemental Experimental Procedures
+
+
+
+
+
+Device Fabrication
+
+
+
+Microfluidic devices were designed using AutoCAD software (Autodesk, Inc.), and the components
+
+tested using COMSOL Multiphysics (COMSOL Inc.). A CAD file is also available in (Data S1).
+
+
+
+Devices were fabricated using a bio-compatible, silicon-based polymer, polydimethylsiloxane (PDMS)
+
+via replica molding using the epoxy-based photo resist SU8 as the master, as previously described
+
+(Mazutis et al., 2013; McDonald et al., 2000). The PDMS devices were then rendered hydrophobic by
+
+flowing in Aquapel (Rider, MA, USA) through the channels, drying out the excess fluid by flowing in
+
+pressurized air, and baking the device at 65ºC for 10 minutes.
+
+
+
+Bead Synthesis
+
+
+
+Bead functionalization and reverse direction phosphoramidite synthesis (5’ to 3’) were performed by
+
+Chemgenes Corp. Toyopearl HW-65S resin (~30 micron mean particle diameter) was purchased from
+
+Tosoh Biosciences (catalog #19815, Tosoh Bioscience), and surface hydroxyls were reacted with a
+
+PEG derivative to generate an 18-carbon long, flexible-chain linker. The functionalized bead was then
+
+used as a solid support for reverse-direction phosphoramidite synthesis (5’3’) on an Expedite 8909
+
+DNA/RNA synthesizer using DNA Synthesis at 10 micromole scale and a coupling time of 3 minutes.
+
+Amidites used were: N6-Benzoyl-3’-O-DMT-2’- deoxyadenosine-5’-cyanoethyl-N,N-diisopropyl-
+
+phosphoramidite (dA-N6-Bz-CEP); N4-Acetyl-3’-O-DMT-2’-deoxycytidine-5’-cyanoethyl-N,N-
+
+diisopropyl-phosphoramidite (dC-N4-Ac-CEP); N2-DMF-3’-O-DMT-2’- deoxyguanosine-5’-
+
+## Page 3
+
+cyanoethyl-N,N-diisopropyl-phosphoramidite (dG-N2-DMF-CEP); and 3’-O-DMT-2’- deoxythymidine-
+
+5’-cyanoethyl-N,N-diisopropyl-phosphoramidite (T-CEP). Acetic anhydride and N-methylimidazole
+
+were used in the capping step; ethylthio-tetrazole was used in the activation step; iodine was used in the
+
+oxidation step, and dichloroacetic acid was used in the deblocking step. After each of the twelve split-
+
+and-pool phosphoramidite synthesis cycles, beads were removed from the synthesis column, pooled,
+
+hand-mixed, and apportioned into four equal portions by mass; these bead aliquots were then placed in
+
+a separate synthesis column and reacted with either dG, dC, dT, or dA phosphoramidite. This process
+
+was repeated 12 times for a total of 4^12 = 16,777,216 unique barcode sequences. For complete details
+
+regarding the barcoded bead sequences used, see Table S6.
+
+
+
+Cell Culture
+
+
+
+Human 293 T cells were purchased from ATCC (cat # CRL-11268); murine NIH/3T3 cells were
+
+purchased from ATCC (cat # CRL-1658).
+
+
+
+293T and 3T3 cells were grown in DMEM purchased from Invitrogen (cat # 11965092) supplemented
+
+with 10% FBS (Life Technologies, cat # 10437-028) and 1% penicillin-streptomycin (cat # 15070-063).
+
+
+
+Cells were grown to a confluence of 30-60% and treated with TrypLE (Invitrogen, cat #12604013) for
+
+five min, quenched with equal volume of growth medium, and spun down at 300 x g for 5 min. The
+
+supernatant was removed, and cells were resuspended in 1 mL of 1x PBS + 0.2% BSA (Sigma cat
+
+#A8806) and re-spun at 300 x g for 3 min. The supernatant was again removed, and the cells re-
+
+suspended in 1 mL of 1x PBS, passed through a 40-micron cell strainer (Falcon, VWR cat #21008-
+
+949), and counted. For Drop-Seq, cells were diluted to the final concentration in 1x PBS + 200 μg/mL
+
+BSA (NEB, cat # B9000S).
+
+## Page 4
+
+Generation of Whole Retina Suspensions
+
+
+
+Single-cell suspensions were prepared from P14 mouse retinas by adapting previously described
+
+methods for purifying retinal ganglion cells from rat retina (Barres et al., 1988). Briefly, mouse retinas
+
+were digested in a papain solution (40U papain / 10mL DPBS) for 45 minutes. Papain was then
+
+neutralized in a trypsin inhibitor solution (0.15% ovomucoid in DPBS) and the tissue was triturated to
+
+generate a single-cell suspension. Following trituration, the cells were pelleted, resuspended, and
+
+filtered through a 20μm Nitex mesh filter to eliminate any clumped cells. The cells were then diluted in
+
+DPBS + 0.2% BSA (Sigma #A8806) to either 200 cells / μL (replicates 1-6) or 30 cells / μL (replicate
+
+7).
+
+
+
+Retina suspensions were processed through Drop-Seq on four separate days. One library was prepared
+
+on day 1 (replicate 1); two libraries on day 2 (replicates 2 and 3); three libraries on day 3 (replicates 4-
+
+6); and one library on day 4 (replicate 7, high purity). To replicates 4-6, human HEK cells were spiked
+
+in at a concentration of 1 cell / μL (0.5%) but the wide range of cell sizes in the retina data made it
+
+impossible to calibrate single-cell purity or doublets by cross-species comparison. Each of the seven
+
+replicates was sequenced separately.
+
+
+
+Experiments were approved by the institutional animal use and care committee at Harvard Medical
+
+School in accordance with NIH guidelines for the humane treatment of animals.
+
+
+
+Drop-Seq
+
+
+
+Preparation of beads
+
+## Page 5
+
+Beads (either Barcoded Bead SeqA or Barcoded Bead SeqB; Table S6 and see note at end of
+
+Supplemental Experimental Procedures) were washed twice with 30 mL of 100% EtOH and twice
+
+with 30 mL of TE/TW (10 mM Tris pH 8.0, 1 mM EDTA, 0.01% Tween). The bead pellet was
+
+resuspended in 10 mL TE/TW and passed through a 100 µm filter (BD Falcon, cat # 352360) into a 50
+
+mL Falcon tube for long-term storage at 4 oC. The stock concentration of beads (in beads/μL) was
+
+assessed using a Fuchs-Rosenthal cell counter purchased from INCYTO (cat # DHC-F01). For Drop-
+
+Seq, an aliquot of beads was removed from the stock tube, washed in 500 μL of Drop-Seq Lysis Buffer
+
+(DLB, 200 mM Tris pH 7.5, 6% Ficoll PM-400, 0.2% Sarkosyl, 20 mM EDTA), then resuspended in
+
+the appropriate volume of DLB + 50 mM DTT for a bead concentration of ~120 beads/μL.
+
+
+
+Droplet Generation
+
+
+
+The two aqueous suspensions—the single-cell suspension and the bead suspension—were loaded into 3
+
+mL plastic syringes (BD cat #309657). To the bead syringe, we added a 6.4 mm magnetic stir disc
+
+(V&P Scientific, VP cat # 782N-6-150). Droplet generation oil (Biorad, cat # 186-4006) was loaded
+
+into a 10 mL plastic syringe (BD #309604). The three syringes were connected to a 125 μm co-flow
+
+device (Figure S2A) by 0.38 mm inner-diameter polyethylene tubing (Scientific Commodities, inc cat
+
+# BB31695-PE/2), and injected using syringe pumps (KD Scientific, Legato 100) at flow rates of 4.1
+
+mL/hr for each aqueous suspension, and 14 mL/hr for the oil, resulting in ~125 m emulsion drops with
+
+a volume of ~1 nanoliter each. For movie generation, the flow was visualized under an optical
+
+microscope (Olympus IX83) at 10x magnification and imaged at ~1000-2000 frames per second using a
+
+FASTCAM SA5 color camera (Photron, Japan). Droplets were collected in 50 mL falcon tubes; the
+
+collection tube was changed out after every 1 mL of combined aqueous flow volume.
+
+## Page 6
+
+During droplet generation, the beads were kept in suspension by continuous, gentle magnetic stirring
+
+(V&P Scientific, cat # VP710D2). The uniformity in droplet size and the occupancy of beads were
+
+evaluated by observing aliquots of droplets under an optical microscope with bright-field illumination;
+
+in each experiment, greater than 95% of the bead-occupied droplets contained a single bead.
+
+
+
+Droplet Breakage
+
+
+
+The oil from the bottom of each aliquot of droplets was removed with a P1000 pipette, after which 30
+
+mL 6X SSC (Life Technologies, cat # 15557-036) at room temperature was added.
+
+
+
+To break droplets, we added 600 L of Perfluoro-1-octanol (Sigma-Aldrich, cat # 370533-25G), and
+
+shook the tube vigorously by hand for about 20 seconds. The tube was then centrifuged for 1 minute at
+
+1000 x g. To reduce the likelihood of annealed mRNAs dissociating from the beads, samples were kept
+
+on ice for the remainder of the breakage protocol. The supernatant was removed to roughly 5 mL
+
+above the oil-aqueous interface, and the beads washed with an additional 30 mL of room temperature
+
+6X SSC, the aqueous layer transferred to a new tube, and centrifuged again. The supernatant was
+
+removed, and the bead pellet transferred to non-stick 1.5 mL microcentrifuge tubes (VWR, cat # 20170-
+
+650). The pellet was then washed twice with 1 mL 6X SSC, and once with 300 L of 5x Maxima H-
+
+RT buffer (EP0751).
+
+
+
+Reverse Transcription and Exonuclease I Treatment
+
+
+
+To a pellet of up to 90,000 beads, 200 L of RT mix was added, where the RT mix contained 1x
+
+Maxima RT buffer, 4% Ficoll PM-400 (GE Healthcare, cat # 17-0300-05), 1 mM dNTPs (Clontech, cat
+
+# 639125), 1 U/L Rnase Inhibitor (Lucigen, cat # 30281-2), 2.5 M Template_Switch_Oligo (Table
+
+## Page 7
+
+S6), and 10 U/L Maxima H- RT (ThermoScientific cat #EP0751). The beads were incubated at room
+
+temperature for 30 minutes, followed by 42 oC for 90 minutes. The beads were then washed once with
+
+1 mL 1x TE + 0.5% Sodium Dodecyl Sulfate (TE/SDS, Sigma cat# L4522), twice with 1 mL TE/TW,
+
+and once with 10 mM Tris pH 7.5. The bead pellet was then resuspended in 200 L of exonuclease I
+
+mix containing 1x Exonuclease I Buffer and 1 U/L Exonuclease I (NEB cat # B0293S), and incubated
+
+at 37 oC for 45 minutes.
+
+
+
+The beads were then washed once with 1 mL TE/SDS, twice with 1 mL TE/TW, once with 1 mL
+
+ddH2O, and resuspended in ddH2O. Bead concentration was determined using a Fuchs-Rosenthal cell
+
+counter. Aliquots of 1000 beads were amplified by PCR in a volume of 50 L using 1x Hifi HotStart
+
+Readymix (Kapa Biosystems, cat #KK2602) and 0.8 M Template_Switch_PCR primer (Table S6).
+
+
+
+The aliquots were thermocycled as follows: 95 oC 3 min; then four cycles of: 98 oC for 20 sec, 65 oC for
+
+45 sec, 72 oC for 3 min; then X cycles of: 98 oC for 20 sec, 67 oC for 20 sec, 72 oC for 3 min; then a
+
+final extension step of 5 min. For the human-mouse experiment using cultured cells, X was 8 cycles;
+
+for the dissociated retina experiment, X was 9 cycles. Pairs of aliquots were pooled together after PCR
+
+and purified with 0.6x Agencourt AMPure XP beads (Beckman Coulter, cat # A63881) according to the
+
+manufacturer’s instructions, and eluted in 10 L of H2O. Aliquots were pooled according to the number
+
+of STAMPs to be sequenced, and the concentration of the pool quantified on a BioAnalyzer High
+
+Sensitivity Chip (Agilent Technologies, cat # 5067-4626).
+
+
+
+Preparation of Drop-Seq cDNA Library for Sequencing
+
+
+
+To prepare 3’-end cDNA fragments for sequencing, four aliquots of 600 pg of cDNA were used as
+
+input in four standard Nextera XT tagmentation reactions (Illumina, cat #FC-131-1096), performed
+
+## Page 8
+
+according to the manufacturer’s instructions except that 200 nM of the custom primers P5_TSO_Hybrid
+
+and Nextera_N701 (Table S6) were used in place of the kit’s provided oligonucleotides. The samples
+
+were then amplified as follows: 95 oC for 30 sec; 11 cycles of 95 oC for 10 sec, 55 oC for 30 sec, 72 oC
+
+for 30 sec; then a final extension step of 72 oC for 5 min.
+
+
+
+Pairs of the 4 aliquots were pooled together, and then purified using 0.6x Agencourt AMPure XP Beads
+
+according to the manufacturer’s instructions, and eluted in 10 L of water. The two 10 L aliquots
+
+were combined together and the concentration determined using a BioAnayzer High Sensitivity Chip.
+
+The average size of sequenced libraries was between 450 and 650 bp.
+
+
+
+The libraries were sequenced on the Illumina NextSeq 500 using 4.67 pM in a volume of 3 mL HT1,
+
+and 3 mL of 0.3 M Read1CustSeqA or Read1CustSeqB (Table S6 and see note at the end of
+
+Supplemental Experimental Procedures) for priming of read 1. Read 1 was 20 bp (bases 1-12 cell
+
+barcode, bases 13-20 UMI); read 2 (paired end) was 50 bp for the human-mouse experiment, and 60 bp
+
+for the retina experiment.
+
+
+
+Species Contamination Experiment
+
+
+
+To determine the origin of off-species contamination of STAMP libraries (Figure S3D), we: (1)
+
+performed Drop-Seq exactly as above (control experiment) with a HEK/3T3 cell suspension mixture of
+
+100 cells / L in concentration; (2) performed the microfluidic co-flow step with HEK and 3T3 cells
+
+separately, each at a concentration of 100 cells / L, and then mixed droplets prior to breakage; and (3)
+
+performed STAMP generation through exonuclease digestion, with the HEK and 3T3 cells separately,
+
+then mixed equal numbers of STAMPs prior to PCR amplification. A single 1000 microparticle aliquot
+
+was amplified for each of the three conditions, then purified and quantified on a BioAnalyzer High
+
+## Page 9
+
+Sensitivity DNA chip. 600 pg of each library was used in a single Nextera Tagmentation reaction as
+
+described above, except that each of the three libraries was individually barcoded with the primers
+
+Nextera_N701 (condition 1), Nextera_N702 (condition 2), or Nextera_N703 (condition 3), and a total
+
+of 12 PCR cycles were used in the Nextera PCR instead of 11. The resulting library was quantified on
+
+a High Sensitivity DNA chip, and each was loaded at a concentration of 8 pM on a single, multiplexed
+
+MiSeq run using 0.5 M Read1CustSeqA as a custom primer for read 1 (see note at end of this section).
+
+
+
+Soluble RNA Experiments
+
+
+
+To quantify the number of primer annealing sites, 20,000 beads were incubated with 10 M of
+
+polyadenylated synthetic RNA (synRNA, Table S6) in 2x SSC for 5 min at room temperature, and
+
+washed three times with 200 L of TE-TW, then resuspended in 10 L of TE-TW. The beads were
+
+then incubated at 65 oC for 5 minutes, and 1 L of supernatant was removed for spectrophotometric
+
+analysis on the Nanodrop 2000. The concentration was compared with beads that had been treated the
+
+same way, except no synRNA was added.
+
+
+
+To determine whether the bead-bound primers were capable of reverse transcription, and to measure the
+
+homogeneity of the cell barcode sequence on the bead surface, beads were washed with TE-TW, and
+
+added at a concentration of 100 / L to the reverse transcriptase mix described above. This mix was
+
+then co-flowed into the standard Drop-Seq 125 m co-flow device with 200 nM SynRNA in 1x PBS +
+
+0.02% BSA. Droplets were collected and incubated at 42 oC for 30 minutes. 150 L of 50 mM EDTA
+
+was added to the emulsion, followed by 12 L of perfluooctanoic acid to break the emulsion. The
+
+beads were washed twice in 1 mL TE-TW, followed by one wash in H2O, then resuspended in TE.
+
+Eleven beads were handpicked under a microscope into a 50 L PCR mix containing 1x Kapa HiFi
+
+Hotstart PCR mastermix, 400 nM P7-TSO_Hybrid, and 400 nM TruSeq_F (Table S6). The PCR
+
+## Page 10
+
+reaction was cycled as follows: 98 oC for 3 min; 12 cycles of: 98 oC for 20 s, 70 oC for 15 s, 72 oC for 1
+
+min; then a final 72 oC incubation for 5 min. The resulting amplicon was purified on a Zymo DNA
+
+Clean and Concentrator 5 column, and run on a BioAnalyzer High Sensitivity Chip to estimate
+
+concentration. The amplicon was then sequenced on an Illumina MiSeq at a final concentration of 6
+
+pM. Read 1, primed using the standard Illumina TruSeq primer, was a 20 bp molecular barcode on the
+
+SynRNA, while Read 2, primed with CustSynRNASeq, contained the 12 bp cell barcode and 8 bp UMI.
+
+
+
+To estimate the efficiency of Drop-Seq, we used a set of external RNAs (ERCC Spike-ins, Life
+
+Technologies #4456740). We diluted the ERCC spike-ins to 0.32% of the stock in 1x PBS + 1 U/L
+
+RNase Inhibitor (Lucigen) + 200 g/ mL BSA (NEB), and used this in place of the cell flow in the
+
+Drop-Seq protocol, so that each bead was incubated with ~100,000 ERCC mRNA molecules per
+
+nanoliter droplet.  Sequence reads were aligned to a dual ERCC-human (hg19) reference, using the
+
+human sequence as “bait,” which dramatically reduced the number of low-quality alignments to ERCC
+
+transcripts reported by STAR compared with alignment to an ERCC-only reference.
+
+
+
+Standard mRNA-Seq and In-Solution Template Switch Amplification
+
+
+
+To compare Drop-Seq average expression data to standard mRNAseq data, we used 1.815 ug of
+
+purified RNA from 3T3 cells, from which we also prepared and sequenced 550 STAMPs. The RNA
+
+was used in the TruSeq Stranded mRNA Sample Preparation kit (Illumina, # RS-122-2101) according
+
+to the manufacturer’s instructions. For NextSeq 500 sequencing, 0.72 pM of Drop-Seq library was
+
+combined with 0.48 pM of the mRNAseq library in a final volume of 3 mL Buffer HT1.
+
+
+
+To compare Drop-Seq average expression data to mRNAseq libraries prepared by a standard, in-
+
+solution template switch amplification approach, 5 ng of the same purified 3T3 RNA used above was
+
+## Page 11
+
+diluted in 2.75 L of H2O. To the RNA, 1 μL of 10 μM UMI_SMARTdT primer was added (Table
+
+S6) and heated to 72 C, followed by incubation at 4 C for 1 min, after which we added 2 μL 20% Ficoll
+
+PM-400, 2 μL 5x RT Buffer (Maxima H- kit), 1 μL 10 mM dNTPs (Clontech), 0.5 μL 50 μM
+
+Template_Switch_Oligo (Table S6), and 0.5 μL Maxima H- RT. The RT was incubated at 42 C for 90
+
+minutes, followed by heat inactivation for 5 min at 85 C. An RNase cocktail (0.5 μL RNase I,
+
+Epicentre N6901K, and 0.5 μL RNase H, Life Tech 18021071) was added to remove the terminal
+
+riboGs from the template switch oligo, and the sample incubated for 30 min at 37 C. Then, 0.4 μL of
+
+100 μM Template_Switch_PCR primer was added, along with 25 μL 2x Kapa Hifi supermix, and 13.6
+
+μL H2O. The sample was cycled as follows: 95 C 3 min; 14 cycles of: 98 C 20 s, 67 C 20 s, and 72 C
+
+3 min; then 72 C 5 min. The samples were purified with 0.6x AMPure XP beads according to the
+
+manufacturer’s instructions, and eluted in 10 μL H2O. 600 pg of amplified cDNA was used as input
+
+into a Nextera XT reaction. 0.6 pM of library was sequenced on a NextSeq 500, multiplexed with three
+
+other samples; Read1CustSeqB was used to prime read 1.
+
+
+
+
+
+Droplet Digital PCR (ddPCR) Experiments
+
+
+
+To quantify the efficiency of Drop-Seq (Figure S4A), 50,000 HEK cells, prepared in an identical
+
+fashion as in Drop-Seq, were pelleted and RNA purified using the Qiagen RNeasy Plus Kit according to
+
+the manufacturer’s protocol. The eluted RNA was diluted to a final concentration of 1 cell-equivalent
+
+per microliter in an RT-ddPCR reaction containing RT-ddPCR supermix (BioRad, # 186-3021), and a
+
+gene primer-probe set. Droplets were produced using BioRad ddPCR droplet generation system, and
+
+thermocycled with the manufacturer’s recommended protocol, and droplet fluorescence analyzed on the
+
+BioRad QX100 droplet reader. Concentrations of RNA and confidence intervals were computed by
+
+BioRad QuantaSoft software. Three replicates of 50,000 HEK cells were purified in parallel, and the
+
+## Page 12
+
+concentration of each gene in each replicate was measured two independent times. The probes (Life
+
+Technologies #4331182) used were: ACTB (hs01060665_g1), B2M (hs00984230_m1), CCNB1
+
+(mm03053893), EEF2 (hs00157330_m1), ENO1 (hs00361415_m1), GAPDH (hs02758991_g1),
+
+PSMB4 (hs01123843_g1), TOP2A (hs01032137_m1), YBX3 (hs01124964_m1), and YWHAH
+
+(hs00607046_m1).
+
+
+
+To estimate the RNA hybridization efficiency of Drop-Seq (Figures S4B and S4C), human brain total
+
+RNA (Life Technologies #AM7962) was diluted to 40 ng / μL in a volume of 20 μL and combined with
+
+20 μL of barcoded primer beads resuspended in Drop-Seq lysis buffer (DLB, composition shown
+
+above) at a concentration of 2,000 beads / μL. The solution was incubated at 15 minutes with rotation,
+
+then spun down and the supernatant transferred to a fresh tube. The beads were washed 3 times with
+
+100 μL of 6x SSC, resuspended in 50 μL H2O, and heated to 72 C for 5 min to elute RNA off the
+
+beads. The elution step was repeated once and the elutions pooled. All steps of the hybridization
+
+(RNA input, hybridization supernatant, three washes, and combined elution) were separately purified
+
+using the Qiagen RNeasy Plus Mini Kit (cat #74134) according to the manufacturers’ instructions.
+
+Various dilutions of the elutions were used in RT-ddPCR reactions with primers and probes for either
+
+ACTB or GAPDH.
+
+
+
+Fluidigm C1 Experiments
+
+
+
+C1 experiments were performed as previously described (Shalek et al., 2014). Briefly, suspensions of
+
+3T3 and HEK cells were stained with calcein violet and calcein orange (Life Technologies) according
+
+to the manufacturer's recommendations, diluted down to a concentration of 250,000 cells per mL, and
+
+mixed 1:1. This cell mixture was then loaded into two medium C1 cell capture chips from Fluidigm and,
+
+after loading, caught cells were visualized and identified using DAPI and TRITC fluorescence. Bright
+
+## Page 13
+
+field images were used to identify ports with > 1 cell (a total of 14 were identified from the two C1
+
+chips used, out of 192 total). After C1-mediated whole transcriptome amplification, libraries were
+
+made using Nextera XT (Illumina), and loaded on a NextSeq 500 at 2.2 pM. Single-read sequencing
+
+(60 bp) was performed to mimic the read structure in DropSeq, and the reads aligned as per below. Ten
+
+of the 192 cells, containing fewer than 100,000 reads per cell, were excluded from analysis.
+
+
+
+Read Alignment and Generation of Digital Expression Data
+
+
+
+Raw sequence data was first filtered to remove all read pairs with a barcode base quality of less than 10.
+
+The second read (50 or 60 bp) was then trimmed at the 5’ end to remove any TSO adapter sequence,
+
+and at the 3’ end to remove polyA tails of length 6 or greater, then aligned to either the mouse (mm10)
+
+genome (retina experiments) or a combined mouse (mm10) –human (hg19) mega-reference (species
+
+mixing experiments), using STAR v2.4.0a with the default settings.
+
+
+
+Uniquely mapped reads were grouped by cell barcode. To digitally count gene transcripts, a list of
+
+UMIs in each gene, within each cell, was assembled, and UMIs within ED = 1 were merged together.
+
+The total number of distinct UMI sequences was counted, and this number was reported as the number
+
+of transcripts of that gene for a given cell.
+
+
+
+To generate the digital expression matrices in this paper, we performed UMI merging at ED=1,
+
+including insertions and deletions. However, a subsequent comparison of UMI edit distance
+
+relationships within and across genes showed that inclusion of indels resulted in excessive merging
+
+(Table S1). For our ERCC sensitivity analysis, we therefore used substitution-only UMI merging, and
+
+plan to also use this approach in future experiments. Without any edit distance correction (or using the
+
+corrective approach described in Islam et al., 2014), we obtained an efficiency estimate of 47% for the
+
+## Page 14
+
+ERCC dataset shown in Figure 3G, though we believe (from the analysis in Table S1) that for our
+
+data, our own correction approach, and the lower capture-rate estimate derived from it, are more
+
+accurate.
+
+
+
+To distinguish cell barcodes arising from STAMPs, rather than those that corresponded to beads never
+
+exposed to cell lysate, we ordered our digital expression matrix by the total number of transcripts per
+
+cell barcode, and plotted the cumulative fraction of all transcripts in the matrix for each successively
+
+smaller cell barcode. Empirically, our data always displays a “knee” at a cell barcode number close to
+
+the estimated number of STAMPs amplified (Figure S3A). All cell barcodes larger than this cutoff
+
+were used in downstream analysis, while the remaining cell barcodes were discarded.
+
+
+
+
+
+Cell Cycle Analysis of HEK and 3T3 Cells
+
+
+
+Gene sets reflecting five phases of the HeLa cell cycle (G1/S, S, G2/M, M and M/G1) were taken from
+
+Whitfield et al. (Whitfield et al., 2002) (Table S2), and refined by examining the correlation between
+
+the expression pattern of each gene and the average expression pattern of all genes in the respective
+
+gene-set, and excluding genes with a low correlation (R<0.3). This step removed genes that were
+
+identified as phase-specific in HeLa cells but did not correlate with that phase in our single-cell data.
+
+The remaining genes in each refined gene-set were highly correlated (not shown). We then averaged the
+
+normalized expression levels (log2(TPM+1)) of the genes in each gene-set to define the phase-specific
+
+scores of each cell. These scores were then subjected to two normalization steps. First, for each phase,
+
+the scores were centered and divided by their standard deviation. Second, the normalized scores of each
+
+cell were centered and normalized.
+
+## Page 15
+
+To order cells according to their progression along the cell cycle, we first compared the pattern of
+
+phase-specific scores of each cell to eight potential patterns along the cell cycle: only G1/S is on, both
+
+G1/S and S, only S, only G2/M, G2/M and M, only M, only M/G1, M/G1 and G1. We also added a
+
+ninth pattern for equal scores of all phases (either all active or all inactive). Each pattern was defined
+
+simply as a vector of ones for active programs and zeros for inactive programs. We then classified the
+
+cells by the defined patterns based on the maximal correlation of the phase-specific scores with these
+
+potential patterns. Importantly, none of the cells were classified to the ninth pattern of equal activity,
+
+while multiple cells were assigned to each of the other patterns. To further order the cells within each
+
+class, we sorted the cells based on their relative correlation with the preceding and succeeding patterns,
+
+thereby smoothing the transitions between classes (Figure 4A).
+
+
+
+To identify cell cycle-regulated genes we used the cell cycle ordering defined above and a sliding
+
+window approach with a window size of 100 cells. We identified the windows with maximal average
+
+expression and minimal average expression for each gene and used a two-sample t-test to assign an
+
+initial p-value for the difference between maximal and minimal windows. A similar analysis was
+
+performed after shuffling the order of cells to generate control p-values that can be used to evaluate
+
+false-discovery rate (FDR). Specifically, we examined for each potential p-value threshold, how many
+
+genes pass that threshold in the cell cycle ordered and in the randomly ordered analyses to assign FDR.
+
+Genes were defined as being previously known to be cell-cycle regulated if they were included in a cell
+
+cycle GO/KEGG/REACTOME gene set, or reported in a recent genome-wide study of gene expression
+
+in synchronized replicating cells (Bar-Joseph et al., 2008).
+
+
+
+
+
+Unsupervised Dimensionality Reduction and Clustering Analysis of Retina Data
+
+## Page 16
+
+P14 mouse retina suspensions were processed through Drop-Seq in seven different replicates on four
+
+separate days, and each sequenced separately. Raw digital expression matrices were generated for the
+
+seven sequencing runs. The inflection points in the cumulative distribution plot, corresponding to the
+
+number of cells in each sample replicate, were: 6,600, 9,000, 6,120, 7,650, 7,650, 8280, and 4000. The
+
+full 49,300 cells were merged together in a single matrix, and normalized by dividing by the total
+
+number of UMIs per cell, then multiplying by 10,000. All calculations and data were then performed in
+
+log space (i.e. ln(transcripts-per-10,000 +1)).
+
+
+
+Initial Downsampling and Identification of Highly Variable Genes
+
+Rod photoreceptors constitute 60-70% of the retinal cell population. Furthermore, they are significantly
+
+smaller than other retinal cell types (Carter-Dawson and LaVail, 1979), and as a result yielded
+
+significantly fewer genes (and higher levels of noise) in our single cell data. In our preliminary
+
+computational experiments, performing unsupervised dimensionality reduction on the full dataset
+
+resulted in representations that were dominated by noisy variation within the numerous rod subset; this
+
+compromised our ability to resolve the heterogeneity within other cell-types that were comparatively
+
+much rarer (e.g. amacrines, microglia). Thus, to increase the power of unsupervised dimensionality
+
+reduction techniques for discovering these types we first downsampled the 49,300-cell dataset to extract
+
+single-cell libraries where 900 or more genes were detected, resulting in a 13,155-cell “training set”.
+
+We reasoned that this “training set” would be enriched for rare cell types that are larger in size at the
+
+expense of “noisy” rod cells. The remaining 36,145 cells (henceforth “projection set”) were then
+
+directly embedded onto the two-dimensional representation learned from the training set (see below).
+
+This enabled us to leverage the full statistical power of our data to define and annotate cell types.
+
+
+
+We first identified the set of genes that was most variable across our training set, after controlling for
+
+the relationship between mean expression and variability. We calculated the mean and a dispersion
+
+## Page 17
+
+measure (variance/mean) for each gene across all 13,155 single cells, and placed genes into 20 bins
+
+based on their average expression. Within each bin, we then z-normalized the dispersion measure of all
+
+genes within the bin, in order to identify outlier genes whose expression values were highly variable
+
+even when compared to genes with similar average expression. We used a z-score cutoff of 1.7 to
+
+identify 384 highly variable genes.
+
+
+
+Principal Components Analysis
+
+We ran Principal Components Analysis (PCA) on our training set as previously described (Shalek et al.,
+
+2013), using the prcomp function in R, after scaling and centering the data along each gene. We used
+
+only the previously identified “highly variable” genes as input to the PCA in order to ensure robust
+
+identification of the primary structures in the data.
+
+
+
+While the number of principal components returned is equal to the number of profiled cells, only a
+
+small fraction of these components explain a statistically significant proportion of the variance, as
+
+compared to a null model. We used two approaches to identify statistically significant PCs for further
+
+analysis: (1) we performed 10000 independent randomizations of the data such that within each
+
+realization, the values along every row (gene) of the scaled expression matrix are randomly permuted.
+
+This operation randomizes the pairwise correlations between genes while leaving the expression
+
+distribution of every gene unchanged. PCA was performed on each of these 10000 “randomized”
+
+datasets. Significant PCs in the un-permuted data were identified as those with larger eigenvalues
+
+compared to the highest eigenvalues across the 10000 randomized datasets (p < 0.01, Bonferroni
+
+corrected). (2) We modified a randomization approach (‘jack straw’) proposed by Chung and Storey
+
+(Chung and Storey, 2014) and which we have previously applied to single-cell RNA-seq data (Shalek et
+
+al., 2014). Briefly, we performed 1,000 PCAs on the input data, but in each analysis, we randomly
+
+‘scrambled’ 1% of the genes to empirically estimate a null distribution of scores for every gene. We
+
+## Page 18
+
+used the joint-null criterion (Leek and Storey, 2011) to identify PCs that had gene scores significantly
+
+different from the respective null distributions (p<0.01, Bonferroni corrected). Both (1) and (2) yielded
+
+32 ‘significant’ PCs. Visual inspection confirmed that none of these PCs was primarily driven by
+
+mitochondrial, housekeeping, or hemoglobin genes. As expected, markers for distinct retinal cell types
+
+were highly represented among the genes with the largest scores (+ve and –ve) along these PCs (Table
+
+S3).
+
+
+
+t-SNE Representation and Post-Hoc Projection of Remaining Cells
+
+
+
+Because canonical markers for different retinal cell types were strongly represented along the
+
+significant PCs (Figure S5), we reasoned that the loadings for individual cells in our training set along
+
+the principal eigenvectors (also “PC subspace representation”) could be used to separate out distinct
+
+cell types in our data. We note that these loadings leverage information from the 384 genes in the PCA,
+
+and therefore are more robust to technical noise than single-cell measurements of individual genes. We
+
+used these PC loadings as input for t-Distributed Stochastic Neighbor Embedding (tSNE) (van der
+
+Maaten and Hinton, 2008), as implemented in the tsne package in R with the “perplexity” parameter set
+
+to 30. The t-SNE procedure returns a two-dimensional embedding of single cells. Cells with similar
+
+expression signatures of genes within our variable set, and therefore similar PC loadings, will likely
+
+localize near each other in the embedding, and hence distinct cell types should form two-dimensional
+
+point clouds across the tSNE map.
+
+
+
+Prior to identifying and annotating the clusters, we projected the remaining 36,145 cells (the projection
+
+set) onto the tSNE map of the training set by the following procedure:
+
+    (1) We projected these cells onto the subspace defined by the significant PCs identified from the
+
+        training set. Briefly, we centered and scaled the 384 x 36,145 expression matrix corresponding
+
+## Page 19
+
+        to the projection set, considering only the highly variable genes; the scaling parameters of the
+
+        training set were used to center and scale each row. We then multiplied the transpose of this
+
+        scaled expression matrix with the 384 x 32 gene scores matrix learned from the training set
+
+      PCA. This yields a PC “loading” for the cells in the projection set along the 32 significant PCs
+
+        learned on the training set.
+
+    (2) Based on its PC loadings, each cell in the projection set was independently embedded on to the
+
+      tSNE map of the training set introduced earlier using a mathematical framework consistent with
+
+        the original tSNE algorithm (Shekhar et al., 2014). We note that while this approach does not
+
+        discover novel clusters outside of the ones identified from the training set, it sharpens the
+
+         distinctions between different clusters by leveraging the statistical power of the full dataset.
+
+       Moreover, the cells are projected based on their PC signatures, not the raw gene expression
+
+        values, which makes our approach more robust against technical noise in individual gene
+
+       measurements.
+
+
+
+See section “Embedding the projection set onto the tSNE map” below for full details.
+
+
+
+   One potential concern with this “post-hoc projection approach” was the possibility that a cell type
+
+    that is completely absent from the training set might be spuriously projected into one of the defined
+
+    clusters. We tested our projection algorithm on a control dataset to explore this possibility, and
+
+    placed stringent conditions to ensure that only cell types adequately represented within the training
+
+    set are projected to avoid spurious assignments (see ‘“Out of sample” projection test’). Using this
+
+    approach, 97% of the cells in the projection set were successfully embedded, resulting in a tSNE
+
+   map consisting of 48296 out of 49300 sequenced cells (Table S7).
+
+
+
+As an additional validation of our approach, we note that the relative frequencies of different cell types
+
+## Page 20
+
+identified after clustering the full data (see below) closely matches estimates in the literature (Table 1).
+
+With the exception of the rods, all the other cell types were enriched at a median value of 2.3X in the
+
+training set compared to their frequency of the full data. This strongly suggests that our downsampling
+
+approach indeed increases the representation of other cell types at the expense of the rod cells, enabling
+
+us to discover PCs that define these cells.
+
+
+
+Density Clustering to Identify Cell-Types
+
+To identify putative cell types on the tSNE map, we used a density clustering approach implemented in
+
+the DBSCAN R package (Ester et al., 1996), initially setting the reachability distance parameter (eps) to
+
+1.0, and removing clusters less than 20 cells, then setting eps to 1.9, and removing clusters less than 50
+
+cells. The first step (eps=1) resulted in an over-partitioning of the data, but enabled us to easily identify
+
+and remove singleton cells that were located along the interfaces of bigger clusters. Following this
+
+"pruning" step, we re-clustered the data with a larger eps value (1.9) to identify a smaller set of 49
+
+clusters involving 44808 cells (91% of our data) with each cluster containing at least 50 cells. This two-
+
+step pruning strategy enabled us to avoid over-partitioning of the data, while at the same time suppress
+
+the co-option of outlier cells into a neighboring cluster. The 49 clusters were further interrogated
+
+through stringent differential expression tests (see below).
+
+
+
+We next examined the 49 total clusters to ensure that our identified clusters truly represented distinct
+
+cellular classifications, as opposed to over-partitioning. We performed a post-hoc test where we
+
+searched for differentially expressed genes (McDavid et al., 2013) between every pair of clusters
+
+(requiring at least 10 genes, each with an average expression difference greater than 1 natural log value
+
+between clusters with a Bonferroni corrected p<0.01). We iteratively merged cluster pairs that did not
+
+satisfy this criterion, starting with the two most related pairs (lowest number of differentially expressed
+
+genes). This process resulted in 10 merged clusters, leaving 39 remaining.
+
+## Page 21
+
+We then computed average gene expression for each of the 39 remaining clusters, and calculated
+
+Euclidean distances between all pairs, using this data as input for complete-linkage hierarchical
+
+clustering and dendrogram assembly. We then compared each of the 39 clusters to the remaining cells
+
+using a likelihood-ratio test (McDavid et al., 2013) to identify marker genes that were differentially
+
+expressed in the cluster.
+
+
+
+Embedding the Projection Set onto the tSNE Map
+
+We used the computational approach in Shekhar et al. (Shekhar et al., 2014) and Berman et al. (Berman
+
+et al., 2014) to project new cells onto an existing tSNE map. First, the expression vector of the cell is
+
+reduced to include only the set of highly variable genes, and subsequently centered and scaled along
+
+each gene using the mean and standard deviation of the gene expression in the training set. This scaled
+
+expression vector z (dimensions 1 x 384) is multiplied with the scores matrix of the genes S
+
+(dimensions 384 x 32), to obtain its “loadings” along the significant PCs u (dimensions 1 x 32). Thus,
+
+                                             𝑢′ = 𝑧′. 𝑆
+
+u (dimensions 1 x 32) denotes the representation of the new cell in the PC subspace identified from the
+
+training set. We note a point of consistency here in that performing the above dot product on a scaled
+
+expression vector of a cell z taken from the training set recovers its correct subspace representation u, as
+
+it ought to be the case.
+
+Given the PC loadings of the cells in the training set {ui} (i=1,2,…Ntrain) and their tSNE coordinates {yi}
+
+(i=1,2,…Ntrain), the task now is to find the tSNE coordinates y’ of the new cell based on its loadings
+
+vector u’. As in the original tSNE framework (van der Maaten and Hinton, 2008), we “locate” the new
+
+cell in the subspace relative to the cells in the training set by computing a set of transition probabilities,
+
+                                                           2
+                                    exp (−𝑑(𝑢′, 𝑢𝑖) ⁄ 2𝜎𝑢′2 )
+                           𝑝(𝑢′|𝑢𝑖) =                           2                         ∑{𝑢𝑖} exp(−𝑑(𝑢′, 𝑢𝑖)2 ⁄ 2𝜎𝑢′ )
+
+## Page 22
+
+Here, d(. , .) represents Euclidean distances, and the the bandwidth σu’ is chosen by a simple binary
+
+search in order to constrain the Shannon entropy associated with 𝑝(𝑢′|𝑢𝑖) to log2(30), where 30
+
+corresponds to the value of the perplexity parameter used in the tSNE embedding of the training set.
+
+Note that σu’ is chosen independently for each cell.
+
+
+
+A corresponding set of transition probabilities in the low dimensional embedding are defined based on
+
+the Student’s t-distribution as,
+
+                                                       −1
+                                     (1 + 𝑑(𝑦′, 𝑦𝑖)2)
+                              𝑞(𝑦′|𝑦𝑖) =
+                          ∑{𝑦𝑖}(1 + 𝑑(𝑦′, 𝑦𝑖)2)−1
+
+where y’ are the coordinates of the new cell that are unknown. We calculate these by minimizing the
+
+Kullback-Leibler divergence between 𝑝(𝑢′|𝑢𝑖) and 𝑞(𝑦′|𝑦𝑖),
+
+                                                          𝑝(𝑢′|𝑢𝑖)
+                             𝑦′ = 𝑎𝑟𝑔𝑚𝑖𝑛 ∑𝑝(𝑢′|𝑢𝑖) log
+                                                         𝑞(𝑦′|𝑦𝑖)
+                                                                           𝑖
+
+This is a non-convex objective function with respect to its arguments, and is minimized using the
+
+Nelder-Mead simplex algorithm, as implemented in the Matlab function fminsearch. This procedure
+
+can be parallelized across all cells in the projection set.
+
+A few notes on the implementation,
+
+    1.  Since this is a post-hoc projection, and 𝑝(𝑢′|𝑢𝑖) is only a relative measure of pairwise
+
+         similarity in that it is always constrained to sum to 1, we wanted to avoid the possibility of new
+
+         cells being embedded on the tSNE map by virtue of their high relative similarity to one or two
+
+        training cells (“short circuiting”). In other words, we chose to project only those cells that were
+
+       drawn from regions of the PC subspace that were well represented in the training set by at least
+
+       a few cells.
+
+       Thus, we retained a cell u’ for projection only if 𝑝(𝑢′|𝑢𝑖) > 𝑝𝑡ℎ𝑟𝑒𝑠 was true for at least Nmin
+
+         cells in the training set (pthres = 5 × 10−3, Nmin = 10). We calibrated the values for pthres and
+
+## Page 23
+
+        Nmin by testing our projection algorithm on cases where the projection set was known to be
+
+       completely different from the training set to ensure that such cells were largely rejected by this
+
+         constraint. (see Section ‘“Out of sample” projection test’)
+
+    2.  For cells that pass the constraint in pt. 1., the initial value of the tSNE coordinate y’0 is set to,
+
+                                            𝑦′0 = ∑𝑝(𝑢′|𝑢𝑖)𝑦𝑖
+                                                                                   𝑖
+
+           i.e. a weighted average of the tSNE coordinates of the training set with the weights set to the
+
+        pairwise similarity in the PC subspace representation.
+
+    3. A cell satisfying the condition in 1. is said to be “successfully projected” to a location y’* when
+
+       a minimum of the KL divergence could be found within the maximum number of iterations.
+
+      However since the program is non-convex and is guaranteed to only find local minima, we
+
+       wanted to explore if a better minima could be found. Briefly, we uniformly sampled points
+
+       from a 25 x 25 grid centered on y’* to check for points where the value of the KL-divergence
+
+      was within 5% of its value at y’* or lower. Whenever this condition was satisfied (< 2%) of the
+
+        time, we re-ran the optimization by setting the new point as the initial value.
+
+
+
+“Out of Sample” Projection Test
+
+In order to test our post-hoc projection method, we conducted the following computational experiment
+
+wherein each of the 39 distinct clusters on the tSNE map was synthetically “removed” from the tSNE
+
+map, and then reprojected cell-by-cell on the tSNE map of the remaining clusters using the procedure
+
+outlined above. Only cells from the training set were used in these calculations.
+
+   Assuming our cluster distinctions are correct, in each of these 39 experiments, the cluster that is
+
+being reprojected represents an “out of sample” cell type. Thus successful assignments of these cells
+
+into one of the remaining 38 clusters would be spurious. For each of the 39 clusters that was removed
+
+and reprojected, we classified the cells into three groups based on the result of the projection method:
+
+## Page 24
+
+    (1) Cells that did not satisfy the condition 1. in the previous section (i.e. did not have a high
+
+         relative similarity to at least Nmin training cells), and therefore “failed” to project.
+
+    (2) Cells that were successfully assigned a tSNE coordinate y’, but that could not be assigned into
+
+       any of the existing clusters according to the condition below.
+
+    (3) Cells that were successfully assigned a tSNE coordinate y’, and which were “wrongly
+
+        assigned” to one of the existing clusters. A cell was assigned to a cluster whose centroid was
+
+        closest to y’ if and only if the distance between y’ and the centroid was smaller than the cluster
+
+        radius (the distance of the farthest point from the centroid).
+
+Encouragingly for all of the 39 “out of sample” projection experiments, only a small fraction of cells
+
+were spuriously assigned to one of the clusters, i.e. satisfied (3) above with the parameters pthres =
+
+5 × 10−3 and Nmin = 10 (Table S7). This gave us confidence that our post-hoc embedding of the
+
+projection set would not spuriously assign distinct cell types into one of the existing clusters.
+
+
+
+Downsampling Analyses of Retina Data
+
+
+
+To generate the 500-cell and 2000-cell downsampled tSNE plots shown in Figure 5F, the largest 500 or
+
+2000 cells were sampled from the high-purity replicate (replicate 7), and used as input for PCA and
+
+tSNE. Two extreme outlier points were removed from the 500-cell tSNE prior to plotting. To generate
+
+the 9,731-cell downsampled tSNE plot, 10,000 cells were randomly sampled from the full dataset, and
+
+the cells expressing transcripts from more than 900 genes were used in principal components analysis
+
+and tSNE; the remaining (smaller) cells were projected onto the tSNE embedding.
+
+
+
+Immunohistochemistry
+
+## Page 25
+
+Wild-type C57 mice or Mito-P mice, which express CFP in nGnG amacrine and Type 1 bipolar cells
+
+(Kay et al., 2011), were euthanized by intraperitoneal injection of pentobarbital. Eyes were fixed in 4%
+
+PFA in PBS on ice for one hour, followed by dissection and post-fixation of retinas for an additional 30
+
+minutes, then rinsed with PBS. Retinas were frozen and sectioned at 20 μm in a cryostat. Sections were
+
+incubated with primary antibodies (chick anti-GFP [Abcam], rabbit anti-PPP1R17 [Atlas], or goat anti-
+
+VSX2 [Santa Cruz]) overnight at 4°C, and with secondary antibodies (Invitrogen and Jackson
+
+ImmunoResearch) for 2 hours at room temperature. Sections were then mounted using Fluoromount G
+
+(Southern Biotech) and viewed with an Olympus FVB confocal microscope.
+
+
+
+Note on Bead Surface Primers and Custom Sequencing Primers
+
+
+
+During the course of experiments for this paper, we used two batches of beads that had two slightly
+
+different primer sequences (Barcoded Bead SeqA and Barcoded Bead SeqB, Table S6). Barcoded
+
+Bead SeqA was used in the human-mouse experiments, and in replicates 1-3 of the retina experiment.
+
+Replicates 4-7 were performed with Barcoded Bead SeqB. To prime read 1 for Drop-Seq libraries
+
+produced using Barcoded Bead SeqA beads, Read1CustSeqA was used; to prime read 2 for Drop-Seq
+
+libraries produced using Barcoded Bead SeqB beads, Read1CustSeqB was used. ChemGenes plans to
+
+manufacture beads harboring the Barcoded Bead SeqB sequence. These beads should be used with
+
+Read1CustSeqB.
+
+
+
+
+
+Additional Notes Regarding Drop-Seq Implementation
+
+
+
+Cell and Bead Concentrations
+
+## Page 26
+
+Our experiments have shown that the cell concentration used in Drop-Seq has a strong, linear
+
+relationship to the purity and doublet rates of the resulting libraries (Figures 3A, 3B, and S3B). Cell
+
+concentration also linearly affects throughput: ~10,000 single-cell libraries can be processed per hour
+
+when cells are used at a final concentration of 100 cells / μL, and ~1,200 can be processed when cells
+
+are used at a final concentration of 12.5 cells / μL. The trade-off between throughput and purity is
+
+likely to affect users differently, depending on the specific scientific questions being asked. Currently,
+
+for our standard experiments, we use a final concentration of 50 cells / μL, tolerating a small percentage
+
+of doubles and cell contaminants, to be able to easily and reliably process 10,000 cells over the course
+
+of a couple of hours. As recommended above, we currently favor loading beads at a concentration of
+
+120 / μL (final concentration in droplets = 60 / μL), which empirically yields a < 5% bead doublet rate.
+
+
+
+Drop-Seq Start-Up Costs
+
+The main pieces of equipment required to implement Drop-Seq are three syringe pumps (KD Legato
+
+100 pumps, list price ~$2,000 each) a standard inverted microscope (Motic AE31, list price ~$1,900),
+
+and a magnetic stirrer (V&P scientific, #710D2, list price ~$1,200). A fast camera (used to monitor
+
+droplet generation in real time) is not necessary for the great majority of users (droplet quality can be
+
+monitored by simply placing 3 μL of droplets in a Fuchs-Rosenthal hemocytometer with 17 μL of
+
+droplet generation oil to dilute the droplets into a single plane of focus).
+
+## Page 27
+
+Table S1. Analysis of edit distance relationships among UMIs, Related to Figure 3
+
+
+     UMI Sampling             % Reduction in UMI counts
+
+                                Substitution-only collapse     Indel and substitution collapse
+
+       Within a gene                68.2%                     76.1%
+
+      Across genes                19.1%                     45.7%
+
+Edit distance relationships among UMIs. For the data in Figure 3G, the sequences of the UMIs for
+
+each ERCC gene detected in each cell barcode were collapsed at an edit distance of 1, including only
+
+substitutions (left column) or with both substitutions and insertions/deletions (right column). A control
+
+UMI set was prepared for each gene, using an equal number of UMIs sampled randomly across all
+
+genes/cells. The table shows the percent of the original UMIs that were collapsed for each condition.
+
+## Page 28
+
+Table S5. Cost Analysis of Drop-Seq, Related to Figure 5
+
+                                                                               Cost for
+Reagents                                         Supplier         Catalog #
+                                                                                10,000 cells ($)
+Microfluidics costs (tubing, syringes,
+                                                 N/A              N/A              35.00
+droplet generation oil, device fabrication)
+DropSeq lysis buffer (Ficoll, Tris, Sarkosyl,
+                                                 N/A              N/A                9.35
+EDTA, DTT)
+Barcoded microparticles                     Chemgenes              N/A             137.20
+Maxima H– Reverse Transcriptase                 Thermo          EP0753              59.15
+dNTP mix                                         Clontech           639125                7.78
+RNase inhibitor                                     Lucigen           30281-2                3.80
+Template switch oligo                                 IDT              N/A                7.60
+Perfluorooctanol                                 Sigma           370533              11.90
+Exonuclease I                              NEB          M0293L                3.84
+KAPA Hifi HotStart ReadyMix           KAPA BioSystems          KK2602             210.00
+Nextera XT DNA sample preparation kit                  Illumina       FC-131-1096             120.80
+Ampure XP beads                      Beckman Coulter           A63882              37.35
+BioAnalyzer High Sensitivity Chips                        Agilent         5067-4626                9.64
+
+                              Total cost:                                                $653.41
+                       Cost per cell:                                                  $0.065
+
+## Page 29
+
+Table S6. Oligonucleotide Sequences Used in This Study
+
+
+          synRNA            rCrCrUrArCrArCrGrArCrGrCrUrCrUrUrCrCrGrArUrCrUrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNr
+                                   BrArArArArArArArArArArArArArArArArArArArArArArArA
+
+      Barcoded Bead SeqA       5’ –Bead–Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACGTJJJJJJJJJJJJNNNNNNNN
+                                   TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3’
+      Barcoded Bead SeqB       5’ –Bead–Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACJJJJJJJJJJJJNNNNNNNN
+                                   TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3’
+     Template_Switch_Oligo   AAGCAGTGGTATCAACGCAGAGTGAATrGrGrG
+          TSO_PCR          AAGCAGTGGTATCAACGCAGAGT
+        P5-TSO_Hybrid       AATGATACGGCGACCACCGAGATCTACACGCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+        Nextera_N701       CAAGCAGAAGACGGCATACGAGATTCGCCTTAGTCTCGTGGGCTCGG
+        Nextera_N702       CAAGCAGAAGACGGCATACGAGATCTAGTACGGTCTCGTGGGCTCGG
+        Nextera_N703       CAAGCAGAAGACGGCATACGAGATTTCTGCCTGTCTCGTGGGCTCGG
+      Read1CustomSeqA     GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTACGT
+      Read1CustomSeqB     GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTAC
+        P7-TSO_Hybrid       CAAGCAGAAGACGGCATACGAGATCGTGATCGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+
+           TruSeq_F          AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
+
+        CustSynRNASeq       CGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGTAC
+
+       UMI_SMARTdT       AAGCAGTGGTATCAACGCAGAGTACNNNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTT
+
+## Page 30
+
+Table S7. “Out-of-Sample” Projection Test
+
+               # Cells in      # failed to                   # Wrongly   % Wrongly
+  Cluster #                                # Projected
+                   Cluster          project                      Assigned       Assigned
+     1           153          153           0            0             0.00
+     2           271          271           0            0             0.00
+     3           201          201           0            0             0.00
+     4            46           46            0            0             0.00
+     5            63           62            1            0             0.00
+     6           173          156           17            9             5.20
+     7           277          272           5            5             1.81
+     8           115          115           0            0             0.00
+     9           275          275           0            0             0.00
+    10           155          153           2            2             1.29
+    11           165          162           3            3             1.82
+    12           175          175           0            0             0.00
+    13           46           40            6            5            10.87
+    14           89           89            0            0             0.00
+    15           52           44            8            6            11.54
+    16           179          179           0            0             0.00
+    17           284          284           0            0             0.00
+    18           64           63            1            1             1.56
+    19           108          107           1            0             0.00
+    20           206          206           0            0             0.00
+    21           154          154           0            0             0.00
+    22           180          180           0            0             0.00
+    23           183          182           1            1             0.55
+    24          3712         3417          295          180            4.85
+    25          1095         1071          24           18            1.64
+    26          1213         1212           1            0             0.00
+    27           323          318           5            4             1.24
+    28           339          330           9            7             2.06
+    29           332          324           8            6             1.81
+    30           447          426           21           18            4.03
+    31           346          340           6            3             0.87
+    32           235          233           2            2             0.85
+    33           453          450           3            3             0.66
+    34           784          784           0            0             0.00
+    35           27           27            0            0             0.00
+    36           43           43            0            0             0.00
+    37           145          139           6            5             3.45
+    38           30           30            0            0             0.00
+    39           17           17            0            0             0.00
+
+
+For each cluster, the “training” cells were removed from the tSNE plot, and then projected onto the
+tSNE. The number of cells that successfully projected into the embedding, and the number of cells that
+were inappropriately incorporated into a different cluster were tabulated.
+
+## Page 31
+
+References
+
+Bar-Joseph, Z., Siegfried, Z., Brandeis, M., Brors, B., Lu, Y., Eils, R., Dynlacht, B.D., and Simon, I. (2008).
+Genome-wide transcriptional analysis of the human cell cycle identifies genes differentially regulated in normal
+and cancer cells. Proceedings of the National Academy of Sciences of the United States of America 105, 955-960.
+Barres, B.A., Silverstein, B.E., Corey, D.P., and Chun, L.L. (1988). Immunological, morphological, and
+electrophysiological variation among retinal ganglion cells purified by panning. Neuron 1, 791-803.
+Berman, G.J., Choi, D.M., Bialek, W., and Shaevitz, J.W. (2014). Mapping the stereotyped behaviour of freely
+moving fruit flies. Journal of the Royal Society, Interface / the Royal Society 11.
+Carter-Dawson, L.D., and LaVail, M.M. (1979). Rods and cones in the mouse retina. I. Structural analysis using
+light and electron microscopy. The Journal of comparative neurology 188, 245-262.
+Chung, N.C., and Storey, J.D. (2014). Statistical Significance of Variables Driving Systematic Variation in High-
+Dimensional Data. Bioinformatics.
+Ester, M., Kriegel, H.P., Sander, J., and Xu, X. (1996). A density-based algorithm for discovering clusters in large
+spatial databases with noise. (Menlo Park, Calif.: AAAI Press).
+Islam, S., Zeisel, A., Joost, S., La Manno, G., Zajac, P., Kasper, M., Lonnerberg, P., and Linnarsson, S. (2014).
+Quantitative single-cell RNA-seq with unique molecular identifiers. Nature methods 11, 163-166.
+Kay, J.N., Voinescu, P.E., Chu, M.W., and Sanes, J.R. (2011). Neurod6 expression defines new retinal amacrine
+cell subtypes and regulates their fate. Nature neuroscience 14, 965-972.
+Leek, J.T., and Storey, J.D. (2011). The joint null criterion for multiple hypothesis tests. Applications in Genetics
+and Molecular Biology 10, 1-22.
+Matz, M.V., Alieva, N.O., Chenchik, A., and Lukyanov, S. (2003). Amplification of cDNA ends using PCR
+suppression effect and step-out PCR. Methods in molecular biology 221, 41-49.
+Mazutis, L., Gilbert, J., Ung, W.L., Weitz, D.A., Griffiths, A.D., and Heyman, J.A. (2013). Single-cell analysis
+and sorting using droplet-based microfluidics. Nature protocols 8, 870-891.
+McDavid, A., Finak, G., Chattopadyay, P.K., Dominguez, M., Lamoreaux, L., Ma, S.S., Roederer, M., and
+Gottardo, R. (2013). Data exploration, quality control and testing in single-cell qPCR-based gene expression
+experiments. Bioinformatics 29, 461-467.
+McDonald, J.C., Duffy, D.C., Anderson, J.R., Chiu, D.T., Wu, H., Schueller, O.J., and Whitesides, G.M. (2000).
+Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis 21, 27-40.
+Picelli, S., Bjorklund, A.K., Faridani, O.R., Sagasser, S., Winberg, G., and Sandberg, R. (2013). Smart-seq2 for
+sensitive full-length transcriptome profiling in single cells. Nature methods 10, 1096-1098.
+Shalek, A.K., Satija, R., Adiconis, X., Gertner, R.S., Gaublomme, J.T., Raychowdhury, R., Schwartz, S., Yosef,
+N., Malboeuf, C., Lu, D., et al. (2013). Single-cell transcriptomics reveals bimodality in expression and splicing
+in immune cells. Nature 498, 236-240.
+Shalek, A.K., Satija, R., Shuga, J., Trombetta, J.J., Gennert, D., Lu, D., Chen, P., Gertner, R.S., Gaublomme, J.T.,
+Yosef, N., et al. (2014). Single-cell RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510,
+363-369.
+Shekhar, K., Brodin, P., Davis, M.M., and Chakraborty, A.K. (2014). Automatic Classification of Cellular
+Expression by Nonlinear Stochastic Embedding (ACCENSE). Proceedings of the National Academy of Sciences
+of the United States of America 111, 202-207.
+van der Maaten, L., and Hinton, G. (2008). Visualizing Data using t-SNE. Journal of Machine Learning Research
+9, 2579-2605.
+Whitfield, M.L., Sherlock, G., Saldanha, A.J., Murray, J.I., Ball, C.A., Alexander, K.E., Matese, J.C., Perou,
+C.M., Hurt, M.M., Brown, P.O., et al. (2002). Identification of genes periodically expressed in the human cell
+cycle and their expression in tumors. Molecular biology of the cell 13, 1977-2000.

drop_seq/drop-seq_supp.pypdf_text.txt

@@ -0,0 +1,1450 @@
+# pypdf text extraction
+source_pdf: drop-seq_supp.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/drop_seq/drop-seq_supp.pdf
+extraction: pypdf: page.extract_text(extraction_mode="layout")
+
+## Page 1
+
+Cell
+Supplemental Information
+Highly Parallel Genome-wide Expression Profiling
+of Individual Cells Using Nanoliter Droplets
+
+Evan Z. Macosko, Anindita Basu, Rahul Satija, James Nemesh, Karthik Shekhar,
+Melissa Goldman, Itay Tirosh, Allison R. Bialas, Nolan Kamitaki, Emily M. Martersteck,
+John J. Trombetta, David A. Weitz, Joshua R. Sanes, Alex K. Shalek, Aviv Regev,
+Steven A. McCarroll
+
+## Page 2
+
+Supplemental Experimental Procedures
+
+
+
+
+
+Device Fabrication
+
+
+
+Microfluidic devices were designed using AutoCAD software (Autodesk, Inc.), and the components
+
+tested using COMSOL Multiphysics (COMSOL Inc.).   A CAD file is also available in (Data S1).
+
+
+
+Devices were fabricated using a bio-compatible, silicon-based polymer, polydimethylsiloxane (PDMS)
+
+via replica molding using the epoxy-based photo resist SU8 as the master, as previously described
+
+(Mazutis et al., 2013; McDonald et al., 2000). The PDMS devices were then rendered hydrophobic by
+
+flowing in Aquapel (Rider, MA, USA) through the channels, drying out the excess fluid by flowing in
+
+pressurized air, and baking the device at 65ºC for 10 minutes.
+
+
+
+Bead Synthesis
+
+
+
+Bead functionalization and reverse direction phosphoramidite synthesis (5’ to 3’) were performed by
+
+Chemgenes Corp.  Toyopearl HW-65S resin (~30 micron mean particle diameter) was purchased from
+
+Tosoh Biosciences (catalog #19815, Tosoh Bioscience), and surface hydroxyls were reacted with a
+
+PEG derivative to generate an 18-carbon long, flexible-chain linker.  The functionalized bead was then
+
+used as a solid support for reverse-direction phosphoramidite synthesis (5’3’) on an Expedite 8909
+
+DNA/RNA synthesizer using DNA Synthesis at 10 micromole scale and a coupling time of 3 minutes.
+Amidites used were: N6-Benzoyl-3’-O-DMT-2’- deoxyadenosine-5’-cyanoethyl-N,N-diisopropyl-
+
+phosphoramidite (dA-N6-Bz-CEP); N4-Acetyl-3’-O-DMT-2’-deoxycytidine-5’-cyanoethyl-N,N-
+
+diisopropyl-phosphoramidite (dC-N4-Ac-CEP); N2-DMF-3’-O-DMT-2’- deoxyguanosine-5’-
+
+## Page 3
+
+cyanoethyl-N,N-diisopropyl-phosphoramidite (dG-N2-DMF-CEP); and 3’-O-DMT-2’- deoxythymidine-
+
+5’-cyanoethyl-N,N-diisopropyl-phosphoramidite (T-CEP).  Acetic anhydride and N-methylimidazole
+
+were used in the capping step; ethylthio-tetrazole was used in the activation step; iodine was used in the
+
+oxidation step, and dichloroacetic acid was used in the deblocking step.  After each of the twelve split-
+
+and-pool phosphoramidite synthesis cycles, beads were removed from the synthesis column, pooled,
+
+hand-mixed, and apportioned into four equal portions by mass; these bead aliquots were then placed in
+
+a separate synthesis column and reacted with either dG, dC, dT, or dA phosphoramidite.  This process
+
+was repeated 12 times for a total of 4^12 = 16,777,216 unique barcode sequences. For complete details
+
+regarding the barcoded bead sequences used, see Table S6.
+
+
+
+Cell Culture
+
+
+
+Human 293 T cells were purchased from ATCC (cat # CRL-11268); murine NIH/3T3 cells were
+
+purchased from ATCC (cat # CRL-1658).
+
+
+
+293T and 3T3 cells were grown in DMEM purchased from Invitrogen (cat # 11965092) supplemented
+
+with 10% FBS (Life Technologies, cat # 10437-028) and 1% penicillin-streptomycin (cat # 15070-063).
+
+
+
+Cells were grown to a confluence of 30-60% and treated with TrypLE (Invitrogen, cat #12604013) for
+
+five min, quenched with equal volume of growth medium, and spun down at 300 x g for 5 min.  The
+
+supernatant was removed, and cells were resuspended in 1 mL of 1x PBS + 0.2% BSA (Sigma cat
+
+#A8806) and re-spun at 300 x g for 3 min.  The supernatant was again removed, and the cells re-
+
+suspended in 1 mL of 1x PBS, passed through a 40-micron cell strainer (Falcon, VWR cat #21008-
+
+949), and counted.  For Drop-Seq, cells were diluted to the final concentration in 1x PBS + 200 μg/mL
+
+BSA (NEB, cat # B9000S).
+
+## Page 4
+
+Generation of Whole Retina Suspensions
+
+
+
+Single-cell suspensions were prepared from P14 mouse retinas by adapting previously described
+
+methods for purifying retinal ganglion cells from rat retina (Barres et al., 1988).  Briefly, mouse retinas
+
+were digested in a papain solution (40U papain / 10mL DPBS) for 45 minutes.  Papain was then
+
+neutralized in a trypsin inhibitor solution (0.15% ovomucoid in DPBS) and the tissue was triturated to
+
+generate a single-cell suspension.  Following trituration, the cells were pelleted, resuspended, and
+
+filtered through a 20μm Nitex mesh filter to eliminate any clumped cells.  The cells were then diluted in
+
+DPBS + 0.2% BSA (Sigma #A8806) to either 200 cells / μL (replicates 1-6) or 30 cells / μL (replicate
+
+7).
+
+
+
+Retina suspensions were processed through Drop-Seq on four separate days.  One library was prepared
+
+on day 1 (replicate 1); two libraries on day 2 (replicates 2 and 3); three libraries on day 3 (replicates 4-
+
+6); and one library on day 4 (replicate 7, high purity).   To replicates 4-6, human HEK cells were spiked
+
+in at a concentration of 1 cell / μL (0.5%) but the wide range of cell sizes in the retina data made it
+
+impossible to calibrate single-cell purity or doublets by cross-species comparison.  Each of the seven
+
+replicates was sequenced separately.
+
+
+
+Experiments were approved by the institutional animal use and care committee at Harvard Medical
+
+School in accordance with NIH guidelines for the humane treatment of animals.
+
+
+
+Drop-Seq
+
+
+
+Preparation of beads
+
+## Page 5
+
+Beads (either Barcoded Bead SeqA or Barcoded Bead SeqB; Table S6 and see note at end of
+
+Supplemental Experimental Procedures) were washed twice with 30 mL of 100% EtOH and twice
+
+with 30 mL of TE/TW (10 mM Tris pH 8.0, 1 mM EDTA, 0.01% Tween). The bead pellet was
+
+resuspended in 10 mL TE/TW and passed through a 100 µm filter (BD Falcon, cat # 352360) into a 50
+mL Falcon tube for long-term storage at 4 oC. The stock concentration of beads (in beads/μL) was
+
+assessed using a Fuchs-Rosenthal cell counter purchased from INCYTO (cat # DHC-F01). For Drop-
+
+Seq, an aliquot of beads was removed from the stock tube, washed in 500 μL of Drop-Seq Lysis Buffer
+
+(DLB, 200 mM Tris pH 7.5, 6% Ficoll PM-400, 0.2% Sarkosyl, 20 mM EDTA), then resuspended in
+
+the appropriate volume of DLB + 50 mM DTT for a bead concentration of ~120 beads/μL.
+
+
+
+Droplet Generation
+
+
+
+The two aqueous suspensions—the single-cell suspension and the bead suspension—were loaded into 3
+
+mL plastic syringes (BD cat #309657).  To the bead syringe, we added a 6.4 mm magnetic stir disc
+
+(V&P Scientific, VP cat # 782N-6-150).  Droplet generation oil (Biorad, cat # 186-4006) was loaded
+
+into a 10 mL plastic syringe (BD #309604).   The three syringes were connected to a 125 μm co-flow
+
+device (Figure S2A) by 0.38 mm inner-diameter polyethylene tubing (Scientific Commodities, inc cat
+
+# BB31695-PE/2), and injected using syringe pumps (KD Scientific, Legato 100) at flow rates of 4.1
+
+mL/hr for each aqueous suspension, and 14 mL/hr for the oil, resulting in ~125 m emulsion drops with
+
+a volume of ~1 nanoliter each. For movie generation, the flow was visualized under an optical
+
+microscope (Olympus IX83) at 10x magnification and imaged at ~1000-2000 frames per second using a
+
+FASTCAM SA5 color camera (Photron, Japan).  Droplets were collected in 50 mL falcon tubes; the
+
+collection tube was changed out after every 1 mL of combined aqueous flow volume.
+
+## Page 6
+
+During droplet generation, the beads were kept in suspension by continuous, gentle magnetic stirring
+
+(V&P Scientific, cat # VP710D2).  The uniformity in droplet size and the occupancy of beads were
+
+evaluated by observing aliquots of droplets under an optical microscope with bright-field illumination;
+
+in each experiment, greater than 95% of the bead-occupied droplets contained a single bead.
+
+
+
+Droplet Breakage
+
+
+
+The oil from the bottom of each aliquot of droplets was removed with a P1000 pipette, after which 30
+
+mL 6X SSC (Life Technologies, cat # 15557-036) at room temperature was added.
+
+
+
+To break droplets, we added 600 L of Perfluoro-1-octanol (Sigma-Aldrich, cat # 370533-25G), and
+
+shook the tube vigorously by hand for about 20 seconds.  The tube was then centrifuged for 1 minute at
+
+1000 x g. To reduce the likelihood of annealed mRNAs dissociating from the beads, samples were kept
+
+on ice for the remainder of the breakage protocol.  The supernatant was removed to roughly 5 mL
+
+above the oil-aqueous interface, and the beads washed with an additional 30 mL of room temperature
+
+6X SSC, the aqueous layer transferred to a new tube, and centrifuged again. The supernatant was
+
+removed, and the bead pellet transferred to non-stick 1.5 mL microcentrifuge tubes (VWR, cat # 20170-
+
+650). The pellet was then washed twice with 1 mL 6X SSC, and once with 300 L of 5x Maxima H-
+
+RT buffer (EP0751).
+
+
+
+Reverse Transcription and Exonuclease I Treatment
+
+
+
+To a pellet of up to 90,000 beads, 200 L of RT mix was added, where the RT mix contained 1x
+
+Maxima RT buffer, 4% Ficoll PM-400 (GE Healthcare, cat # 17-0300-05), 1 mM dNTPs (Clontech, cat
+
+# 639125), 1 U/L Rnase Inhibitor (Lucigen, cat # 30281-2), 2.5 M Template_Switch_Oligo (Table
+
+## Page 7
+
+S6), and 10 U/L Maxima H- RT (ThermoScientific cat #EP0751).  The beads were incubated at room
+temperature for 30 minutes, followed by 42 oC for 90 minutes.  The beads were then washed once with
+
+1 mL 1x TE + 0.5% Sodium Dodecyl Sulfate (TE/SDS, Sigma cat# L4522), twice with 1 mL TE/TW,
+
+and once with 10 mM Tris pH 7.5.  The bead pellet was then resuspended in 200 L of exonuclease I
+
+mix containing 1x Exonuclease I Buffer and 1 U/L Exonuclease I (NEB cat # B0293S), and incubated
+at 37 oC for 45 minutes.
+
+
+
+The beads were then washed once with 1 mL TE/SDS, twice with 1 mL TE/TW, once with 1 mL
+
+ddH2O, and resuspended in ddH2O.  Bead concentration was determined using a Fuchs-Rosenthal cell
+
+counter.  Aliquots of 1000 beads were amplified by PCR in a volume of 50 L using 1x Hifi HotStart
+
+Readymix (Kapa Biosystems, cat #KK2602) and 0.8 M Template_Switch_PCR primer (Table S6).
+
+
+The aliquots were thermocycled as follows: 95 oC 3 min; then four cycles of: 98 oC for 20 sec, 65 oC for
+
+45 sec, 72 oC for 3 min; then X cycles of: 98 oC for 20 sec, 67 oC for 20 sec, 72 oC for 3 min; then a
+
+final extension step of 5 min.  For the human-mouse experiment using cultured cells, X was 8 cycles;
+
+for the dissociated retina experiment, X was 9 cycles.  Pairs of aliquots were pooled together after PCR
+
+and purified with 0.6x Agencourt AMPure XP beads (Beckman Coulter, cat # A63881) according to the
+
+manufacturer’s instructions, and eluted in 10 L of H2O. Aliquots were pooled according to the number
+
+of STAMPs to be sequenced, and the concentration of the pool quantified on a BioAnalyzer High
+
+Sensitivity Chip (Agilent Technologies, cat # 5067-4626).
+
+
+
+Preparation of Drop-Seq cDNA Library for Sequencing
+
+
+
+To prepare 3’-end cDNA fragments for sequencing, four aliquots of 600 pg of cDNA were used as
+
+input in four standard Nextera XT tagmentation reactions (Illumina, cat #FC-131-1096), performed
+
+## Page 8
+
+according to the manufacturer’s instructions except that 200 nM of the custom primers P5_TSO_Hybrid
+
+and Nextera_N701 (Table S6) were used in place of the kit’s provided oligonucleotides.  The samples
+were then amplified as follows:  95 oC for 30 sec; 11 cycles of 95 oC for 10 sec, 55 oC for 30 sec, 72 oC
+
+for 30 sec; then a final extension step of 72 oC for 5 min.
+
+
+
+Pairs of the 4 aliquots were pooled together, and then purified using 0.6x Agencourt AMPure XP Beads
+
+according to the manufacturer’s instructions, and eluted in 10 L of water.  The two 10 L aliquots
+
+were combined together and the concentration determined using a BioAnayzer High Sensitivity Chip.
+
+The average size of sequenced libraries was between 450 and 650 bp.
+
+
+
+The libraries were sequenced on the Illumina NextSeq 500 using 4.67 pM in a volume of 3 mL HT1,
+
+and 3 mL of 0.3 M Read1CustSeqA or Read1CustSeqB (Table S6 and see note at the end of
+
+Supplemental Experimental Procedures) for priming of read 1.  Read 1 was 20 bp (bases 1-12 cell
+
+barcode, bases 13-20 UMI); read 2 (paired end) was 50 bp for the human-mouse experiment, and 60 bp
+
+for the retina experiment.
+
+
+
+Species Contamination Experiment
+
+
+
+To determine the origin of off-species contamination of STAMP libraries (Figure S3D), we: (1)
+
+performed Drop-Seq exactly as above (control experiment) with a HEK/3T3 cell suspension mixture of
+
+100 cells / L in concentration; (2) performed the microfluidic co-flow step with HEK and 3T3 cells
+
+separately, each at a concentration of 100 cells / L, and then mixed droplets prior to breakage; and (3)
+
+performed STAMP generation through exonuclease digestion, with the HEK and 3T3 cells separately,
+
+then mixed equal numbers of STAMPs prior to PCR amplification.  A single 1000 microparticle aliquot
+
+was amplified for each of the three conditions, then purified and quantified on a BioAnalyzer High
+
+## Page 9
+
+Sensitivity DNA chip.  600 pg of each library was used in a single Nextera Tagmentation reaction as
+
+described above, except that each of the three libraries was individually barcoded with the primers
+
+Nextera_N701 (condition 1), Nextera_N702 (condition 2), or Nextera_N703 (condition 3), and a total
+
+of 12 PCR cycles were used in the Nextera PCR instead of 11.  The resulting library was quantified on
+
+a High Sensitivity DNA chip, and each was loaded at a concentration of 8 pM on a single, multiplexed
+
+MiSeq run using 0.5 M Read1CustSeqA as a custom primer for read 1 (see note at end of this section).
+
+
+
+Soluble RNA Experiments
+
+
+
+To quantify the number of primer annealing sites, 20,000 beads were incubated with 10 M of
+
+polyadenylated synthetic RNA (synRNA, Table S6) in 2x SSC for 5 min at room temperature, and
+
+washed three times with 200 L of TE-TW, then resuspended in 10 L of TE-TW.  The beads were
+then incubated at 65 oC for 5 minutes, and 1 L of supernatant was removed for spectrophotometric
+
+analysis on the Nanodrop 2000.  The concentration was compared with beads that had been treated the
+
+same way, except no synRNA was added.
+
+
+
+To determine whether the bead-bound primers were capable of reverse transcription, and to measure the
+
+homogeneity of the cell barcode sequence on the bead surface, beads were washed with TE-TW, and
+
+added at a concentration of 100 / L to the reverse transcriptase mix described above.  This mix was
+
+then co-flowed into the standard Drop-Seq 125 m co-flow device with 200 nM SynRNA in 1x PBS +
+0.02% BSA.  Droplets were collected and incubated at 42 oC for 30 minutes.  150 L of 50 mM EDTA
+
+was added to the emulsion, followed by 12 L of perfluooctanoic acid to break the emulsion.  The
+
+beads were washed twice in 1 mL TE-TW, followed by one wash in H2O, then resuspended in TE.
+
+Eleven beads were handpicked under a microscope into a 50 L PCR mix containing 1x Kapa HiFi
+
+Hotstart PCR mastermix, 400 nM P7-TSO_Hybrid, and 400 nM TruSeq_F (Table S6).  The PCR
+
+## Page 10
+
+reaction was cycled as follows: 98 oC for 3 min; 12 cycles of: 98 oC for 20 s, 70 oC for 15 s, 72 oC for 1
+
+min; then a final 72 oC incubation for 5 min.  The resulting amplicon was purified on a Zymo DNA
+
+Clean and Concentrator 5 column, and run on a BioAnalyzer High Sensitivity Chip to estimate
+
+concentration.  The amplicon was then sequenced on an Illumina MiSeq at a final concentration of 6
+
+pM.  Read 1, primed using the standard Illumina TruSeq primer, was a 20 bp molecular barcode on the
+
+SynRNA, while Read 2, primed with CustSynRNASeq, contained the 12 bp cell barcode and 8 bp UMI.
+
+
+
+To estimate the efficiency of Drop-Seq, we used a set of external RNAs (ERCC Spike-ins, Life
+
+Technologies #4456740).  We diluted the ERCC spike-ins to 0.32% of the stock in 1x PBS + 1 U/L
+
+RNase Inhibitor (Lucigen) + 200 g/ mL BSA (NEB), and used this in place of the cell flow in the
+
+Drop-Seq protocol, so that each bead was incubated with ~100,000 ERCC mRNA molecules per
+
+nanoliter droplet.   Sequence reads were aligned to a dual ERCC-human (hg19) reference, using the
+
+human sequence as “bait,” which dramatically reduced the number of low-quality alignments to ERCC
+
+transcripts reported by STAR compared with alignment to an ERCC-only reference.
+
+
+
+Standard mRNA-Seq and In-Solution Template Switch Amplification
+
+
+
+To compare Drop-Seq average expression data to standard mRNAseq data, we used 1.815 ug of
+
+purified RNA from 3T3 cells, from which we also prepared and sequenced 550 STAMPs.  The RNA
+
+was used in the TruSeq Stranded mRNA Sample Preparation kit (Illumina, # RS-122-2101) according
+
+to the manufacturer’s instructions.  For NextSeq 500 sequencing, 0.72 pM of Drop-Seq library was
+
+combined with 0.48 pM of the mRNAseq library in a final volume of 3 mL Buffer HT1.
+
+
+
+To compare Drop-Seq average expression data to mRNAseq libraries prepared by a standard, in-
+
+solution template switch amplification approach, 5 ng of the same purified 3T3 RNA used above was
+
+## Page 11
+
+diluted in 2.75 L of H2O.  To the RNA, 1 μL of 10 μM UMI_SMARTdT primer was added (Table
+
+S6) and heated to 72 C, followed by incubation at 4 C for 1 min, after which we added 2 μL 20% Ficoll
+
+PM-400, 2 μL 5x RT Buffer  (Maxima H- kit), 1 μL 10 mM dNTPs (Clontech), 0.5 μL 50 μM
+
+Template_Switch_Oligo (Table S6), and 0.5 μL Maxima H- RT.  The RT was incubated at 42 C for 90
+
+minutes, followed by heat inactivation for 5 min at 85 C.  An RNase cocktail (0.5 μL RNase I,
+
+Epicentre N6901K, and 0.5 μL RNase H, Life Tech 18021071) was added to remove the terminal
+
+riboGs from the template switch oligo, and the sample incubated for 30 min at 37 C.  Then, 0.4 μL of
+
+100 μM Template_Switch_PCR primer was added, along with 25 μL 2x Kapa Hifi supermix, and 13.6
+
+μL H2O.  The sample was cycled as follows:  95 C 3 min; 14 cycles of: 98 C 20 s, 67 C 20 s, and 72 C
+
+3 min; then 72 C 5 min.  The samples were purified with 0.6x AMPure XP beads according to the
+
+manufacturer’s instructions, and eluted in 10 μL H2O.  600 pg of amplified cDNA was used as input
+
+into a Nextera XT reaction.  0.6 pM of library was sequenced on a NextSeq 500, multiplexed with three
+
+other samples; Read1CustSeqB was used to prime read 1.
+
+
+
+
+
+Droplet Digital PCR (ddPCR) Experiments
+
+
+
+To quantify the efficiency of Drop-Seq (Figure S4A), 50,000 HEK cells, prepared in an identical
+
+fashion as in Drop-Seq, were pelleted and RNA purified using the Qiagen RNeasy Plus Kit according to
+
+the manufacturer’s protocol.  The eluted RNA was diluted to a final concentration of 1 cell-equivalent
+
+per microliter in an RT-ddPCR reaction containing RT-ddPCR supermix (BioRad, # 186-3021), and a
+
+gene primer-probe set.  Droplets were produced using BioRad ddPCR droplet generation system, and
+
+thermocycled with the manufacturer’s recommended protocol, and droplet fluorescence analyzed on the
+
+BioRad QX100 droplet reader.  Concentrations of RNA and confidence intervals were computed by
+
+BioRad QuantaSoft software.  Three replicates of 50,000 HEK cells were purified in parallel, and the
+
+## Page 12
+
+concentration of each gene in each replicate was measured two independent times.  The probes (Life
+
+Technologies #4331182) used were: ACTB (hs01060665_g1), B2M (hs00984230_m1), CCNB1
+
+(mm03053893), EEF2 (hs00157330_m1), ENO1 (hs00361415_m1), GAPDH (hs02758991_g1),
+
+PSMB4 (hs01123843_g1), TOP2A (hs01032137_m1), YBX3 (hs01124964_m1), and YWHAH
+
+(hs00607046_m1).
+
+
+
+To estimate the RNA hybridization efficiency of Drop-Seq (Figures S4B and S4C), human brain total
+
+RNA (Life Technologies #AM7962) was diluted to 40 ng / μL in a volume of 20 μL and combined with
+
+20 μL of barcoded primer beads resuspended in Drop-Seq lysis buffer (DLB, composition shown
+
+above) at a concentration of 2,000 beads / μL.  The solution was incubated at 15 minutes with rotation,
+
+then spun down and the supernatant transferred to a fresh tube.  The beads were washed 3 times with
+
+100 μL of 6x SSC, resuspended in 50 μL H2O, and heated to 72 C for 5 min to elute RNA off the
+
+beads.  The elution step was repeated once and the elutions pooled.  All steps of the hybridization
+
+(RNA input, hybridization supernatant, three washes, and combined elution) were separately purified
+
+using the Qiagen RNeasy Plus Mini Kit (cat #74134) according to the manufacturers’ instructions.
+
+Various dilutions of the elutions were used in RT-ddPCR reactions with primers and probes for either
+
+ACTB or GAPDH.
+
+
+
+Fluidigm C1 Experiments
+
+
+
+C1 experiments were performed as previously described (Shalek et al., 2014). Briefly, suspensions of
+
+3T3 and HEK cells were stained with calcein violet and calcein orange (Life Technologies) according
+
+to the manufacturer's recommendations, diluted down to a concentration of 250,000 cells per mL, and
+
+mixed 1:1. This cell mixture was then loaded into two medium C1 cell capture chips from Fluidigm and,
+
+after loading, caught cells were visualized and identified using DAPI and TRITC fluorescence. Bright
+
+## Page 13
+
+field images were used to identify ports with > 1 cell (a total of 14 were identified from the two C1
+
+chips used, out of 192 total).  After C1-mediated whole transcriptome amplification, libraries were
+
+made using Nextera XT (Illumina), and loaded on a NextSeq 500 at 2.2 pM.  Single-read sequencing
+
+(60 bp) was performed to mimic the read structure in DropSeq, and the reads aligned as per below.  Ten
+
+of the 192 cells, containing fewer than 100,000 reads per cell, were excluded from analysis.
+
+
+
+Read Alignment and Generation of Digital Expression Data
+
+
+
+Raw sequence data was first filtered to remove all read pairs with a barcode base quality of less than 10.
+
+The second read (50 or 60 bp) was then trimmed at the 5’ end to remove any TSO adapter sequence,
+
+and at the 3’ end to remove polyA tails of length 6 or greater, then aligned to either the mouse (mm10)
+
+genome (retina experiments) or a combined mouse (mm10) –human (hg19) mega-reference (species
+
+mixing experiments), using STAR v2.4.0a with the default settings.
+
+
+
+Uniquely mapped reads were grouped by cell barcode.  To digitally count gene transcripts, a list of
+
+UMIs in each gene, within each cell, was assembled, and UMIs within ED = 1 were merged together.
+
+The total number of distinct UMI sequences was counted, and this number was reported as the number
+
+of transcripts of that gene for a given cell.
+
+
+
+To generate the digital expression matrices in this paper, we performed UMI merging at ED=1,
+
+including insertions and deletions.  However, a subsequent comparison of UMI edit distance
+
+relationships within and across genes showed that inclusion of indels resulted in excessive merging
+
+(Table S1).  For our ERCC sensitivity analysis, we therefore used substitution-only UMI merging, and
+
+plan to also use this approach in future experiments. Without any edit distance correction (or using the
+
+corrective approach described in Islam et al., 2014), we obtained an efficiency estimate of 47% for the
+
+## Page 14
+
+ERCC dataset shown in Figure 3G, though we believe (from the analysis in Table S1) that for our
+
+data, our own correction approach, and the lower capture-rate estimate derived from it, are more
+
+accurate.
+
+
+
+To distinguish cell barcodes arising from STAMPs, rather than those that corresponded to beads never
+
+exposed to cell lysate, we ordered our digital expression matrix by the total number of transcripts per
+
+cell barcode, and plotted the cumulative fraction of all transcripts in the matrix for each successively
+
+smaller cell barcode.  Empirically, our data always displays a “knee” at a cell barcode number close to
+
+the estimated number of STAMPs amplified (Figure S3A).  All cell barcodes larger than this cutoff
+
+were used in downstream analysis, while the remaining cell barcodes were discarded.
+
+
+
+
+
+Cell Cycle Analysis of HEK and 3T3 Cells
+
+
+
+Gene sets reflecting five phases of the HeLa cell cycle (G1/S, S, G2/M, M and M/G1) were taken from
+
+Whitfield et al. (Whitfield et al., 2002) (Table S2), and refined by examining the correlation between
+
+the expression pattern of each gene and the average expression pattern of all genes in the respective
+
+gene-set, and excluding genes with a low correlation (R<0.3). This step removed genes that were
+
+identified as phase-specific in HeLa cells but did not correlate with that phase in our single-cell data.
+
+The remaining genes in each refined gene-set were highly correlated (not shown). We then averaged the
+
+normalized expression levels (log2(TPM+1)) of the genes in each gene-set to define the phase-specific
+
+scores of each cell. These scores were then subjected to two normalization steps. First, for each phase,
+
+the scores were centered and divided by their standard deviation. Second, the normalized scores of each
+
+cell were centered and normalized.
+
+## Page 15
+
+To order cells according to their progression along the cell cycle, we first compared the pattern of
+
+phase-specific scores of each cell to eight potential patterns along the cell cycle: only G1/S is on, both
+
+G1/S and S, only S, only G2/M, G2/M and M, only M, only M/G1, M/G1 and G1. We also added a
+
+ninth pattern for equal scores of all phases (either all active or all inactive). Each pattern was defined
+
+simply as a vector of ones for active programs and zeros for inactive programs. We then classified the
+
+cells by the defined patterns based on the maximal correlation of the phase-specific scores with these
+
+potential patterns. Importantly, none of the cells were classified to the ninth pattern of equal activity,
+
+while multiple cells were assigned to each of the other patterns. To further order the cells within each
+
+class, we sorted the cells based on their relative correlation with the preceding and succeeding patterns,
+
+thereby smoothing the transitions between classes (Figure 4A).
+
+
+
+To identify cell cycle-regulated genes we used the cell cycle ordering defined above and a sliding
+
+window approach with a window size of 100 cells. We identified the windows with maximal average
+
+expression and minimal average expression for each gene and used a two-sample t-test to assign an
+
+initial p-value for the difference between maximal and minimal windows. A similar analysis was
+
+performed after shuffling the order of cells to generate control p-values that can be used to evaluate
+
+false-discovery rate (FDR). Specifically, we examined for each potential p-value threshold, how many
+
+genes pass that threshold in the cell cycle ordered and in the randomly ordered analyses to assign FDR.
+
+Genes were defined as being previously known to be cell-cycle regulated if they were included in a cell
+
+cycle GO/KEGG/REACTOME gene set, or reported in a recent genome-wide study of gene expression
+
+in synchronized replicating cells (Bar-Joseph et al., 2008).
+
+
+
+
+
+Unsupervised Dimensionality Reduction and Clustering Analysis of Retina Data
+
+## Page 16
+
+P14 mouse retina suspensions were processed through Drop-Seq in seven different replicates on four
+
+separate days, and each sequenced separately.  Raw digital expression matrices were generated for the
+
+seven sequencing runs.  The inflection points in the cumulative distribution plot, corresponding to the
+
+number of cells in each sample replicate, were: 6,600, 9,000, 6,120, 7,650, 7,650, 8280, and 4000. The
+
+full 49,300 cells were merged together in a single matrix, and normalized by dividing by the total
+
+number of UMIs per cell, then multiplying by 10,000.  All calculations and data were then performed in
+
+log space (i.e. ln(transcripts-per-10,000 +1)).
+
+
+
+Initial Downsampling and Identification of Highly Variable Genes
+
+Rod photoreceptors constitute 60-70% of the retinal cell population. Furthermore, they are significantly
+
+smaller than other retinal cell types (Carter-Dawson and LaVail, 1979), and as a result yielded
+
+significantly fewer genes (and higher levels of noise) in our single cell data. In our preliminary
+
+computational experiments, performing unsupervised dimensionality reduction on the full dataset
+
+resulted in representations that were dominated by noisy variation within the numerous rod subset; this
+
+compromised our ability to resolve the heterogeneity within other cell-types that were comparatively
+
+much rarer (e.g. amacrines, microglia). Thus, to increase the power of unsupervised dimensionality
+
+reduction techniques for discovering these types we first downsampled the 49,300-cell dataset to extract
+
+single-cell libraries where 900 or more genes were detected, resulting in a 13,155-cell “training set”.
+
+We reasoned that this “training set” would be enriched for rare cell types that are larger in size at the
+
+expense of “noisy” rod cells.  The remaining 36,145 cells (henceforth “projection set”) were then
+
+directly embedded onto the two-dimensional representation learned from the training set (see below).
+
+This enabled us to leverage the full statistical power of our data to define and annotate cell types.
+
+
+
+We first identified the set of genes that was most variable across our training set, after controlling for
+
+the relationship between mean expression and variability. We calculated the mean and a dispersion
+
+## Page 17
+
+measure (variance/mean) for each gene across all 13,155 single cells, and placed genes into 20 bins
+
+based on their average expression.  Within each bin, we then z-normalized the dispersion measure of all
+
+genes within the bin, in order to identify outlier genes whose expression values were highly variable
+
+even when compared to genes with similar average expression. We used a z-score cutoff of 1.7 to
+
+identify 384 highly variable genes.
+
+
+
+Principal Components Analysis
+
+We ran Principal Components Analysis (PCA) on our training set as previously described (Shalek et al.,
+
+2013), using the prcomp function in R, after scaling and centering the data along each gene. We used
+
+only the previously identified “highly variable” genes as input to the PCA in order to ensure robust
+
+identification of the primary structures in the data.
+
+
+
+While the number of principal components returned is equal to the number of profiled cells, only a
+
+small fraction of these components explain a statistically significant proportion of the variance, as
+
+compared to a null model. We used two approaches to identify statistically significant PCs for further
+
+analysis: (1) we performed 10000 independent randomizations of the data such that within each
+
+realization, the values along every row (gene) of the scaled expression matrix are randomly permuted.
+
+This operation randomizes the pairwise correlations between genes while leaving the expression
+
+distribution of every gene unchanged.  PCA was performed on each of these 10000 “randomized”
+
+datasets. Significant PCs in the un-permuted data were identified as those with larger eigenvalues
+
+compared to the highest eigenvalues across the 10000 randomized datasets (p < 0.01, Bonferroni
+
+corrected). (2) We modified a randomization approach (‘jack straw’) proposed by Chung and Storey
+
+(Chung and Storey, 2014) and which we have previously applied to single-cell RNA-seq data (Shalek et
+
+al., 2014).  Briefly, we performed 1,000 PCAs on the input data, but in each analysis, we randomly
+
+‘scrambled’ 1% of the genes to empirically estimate a null distribution of scores for every gene. We
+
+## Page 18
+
+used the joint-null criterion (Leek and Storey, 2011) to identify PCs that had gene scores significantly
+
+different from the respective null distributions (p<0.01, Bonferroni corrected). Both (1) and (2) yielded
+
+32 ‘significant’ PCs. Visual inspection confirmed that none of these PCs was primarily driven by
+
+mitochondrial, housekeeping, or hemoglobin genes.  As expected, markers for distinct retinal cell types
+
+were highly represented among the genes with the largest scores (+ve and –ve) along these PCs (Table
+
+S3).
+
+
+
+t-SNE Representation and Post-Hoc Projection of Remaining Cells
+
+
+
+Because canonical markers for different retinal cell types were strongly represented along the
+
+significant PCs (Figure S5), we reasoned that the loadings for individual cells in our training set along
+
+the principal eigenvectors (also “PC subspace representation”) could be used to separate out distinct
+
+cell types in our data. We note that these loadings leverage information from the 384 genes in the PCA,
+
+and therefore are more robust to technical noise than single-cell measurements of individual genes. We
+
+used these PC loadings as input for t-Distributed Stochastic Neighbor Embedding (tSNE) (van der
+
+Maaten and Hinton, 2008), as implemented in the tsne package in R with the “perplexity” parameter set
+
+to 30. The t-SNE procedure returns a two-dimensional embedding of single cells. Cells with similar
+
+expression signatures of genes within our variable set, and therefore similar PC loadings, will likely
+
+localize near each other in the embedding, and hence distinct cell types should form two-dimensional
+
+point clouds across the tSNE map.
+
+
+
+Prior to identifying and annotating the clusters, we projected the remaining 36,145 cells (the projection
+
+set) onto the tSNE map of the training set by the following procedure:
+
+     (1)  We projected these cells onto the subspace defined by the significant PCs identified from the
+
+          training set. Briefly, we centered and scaled the 384 x 36,145 expression matrix corresponding
+
+## Page 19
+
+          to the projection set, considering only the highly variable genes; the scaling parameters of the
+
+          training set were used to center and scale each row. We then multiplied the transpose of this
+
+          scaled expression matrix with the 384 x 32 gene scores matrix learned from the training set
+
+          PCA. This yields a PC “loading” for the cells in the projection set along the 32 significant PCs
+
+          learned on the training set.
+
+     (2)  Based on its PC loadings, each cell in the projection set was independently embedded on to the
+
+          tSNE map of the training set introduced earlier using a mathematical framework consistent with
+
+          the original tSNE algorithm (Shekhar et al., 2014).  We note that while this approach does not
+
+          discover novel clusters outside of the ones identified from the training set, it sharpens the
+
+          distinctions between different clusters by leveraging the statistical power of the full dataset.
+
+          Moreover, the cells are projected based on their PC signatures, not the raw gene expression
+
+          values, which makes our approach more robust against technical noise in individual gene
+
+          measurements.
+
+
+
+See section  “Embedding the projection set onto the tSNE map” below for full details.
+
+
+
+     One potential concern with this “post-hoc projection approach” was the possibility that a cell type
+
+     that is completely absent from the training set might be spuriously projected into one of the defined
+
+     clusters. We tested our projection algorithm on a control dataset to explore this possibility, and
+
+     placed stringent conditions to ensure that only cell types adequately represented within the training
+
+     set are projected to avoid spurious assignments (see ‘“Out of sample” projection test’).  Using this
+
+     approach, 97% of the cells in the projection set were successfully embedded, resulting in a tSNE
+
+     map consisting of 48296 out of 49300 sequenced cells (Table S7).
+
+
+
+As an additional validation of our approach, we note that the relative frequencies of different cell types
+
+## Page 20
+
+identified after clustering the full data (see below) closely matches estimates in the literature (Table 1).
+
+With the exception of the rods, all the other cell types were enriched at a median value of 2.3X in the
+
+training set compared to their frequency of the full data. This strongly suggests that our downsampling
+
+approach indeed increases the representation of other cell types at the expense of the rod cells, enabling
+
+us to discover PCs that define these cells.
+
+
+
+Density Clustering to Identify Cell-Types
+
+To identify putative cell types on the tSNE map, we used a density clustering approach implemented in
+
+the DBSCAN R package (Ester et al., 1996), initially setting the reachability distance parameter (eps) to
+
+1.0, and removing clusters less than 20 cells, then setting eps to 1.9, and removing clusters less than 50
+
+cells.  The first step (eps=1) resulted in an over-partitioning of the data, but enabled us to easily identify
+
+and remove singleton cells that were located along the interfaces of bigger clusters. Following this
+
+"pruning" step, we re-clustered the data with a larger eps value (1.9) to identify a smaller set of 49
+
+clusters involving 44808 cells (91% of our data) with each cluster containing at least 50 cells. This two-
+
+step pruning strategy enabled us to avoid over-partitioning of the data, while at the same time suppress
+
+the co-option of outlier cells into a neighboring cluster. The 49 clusters were further interrogated
+
+through stringent differential expression tests (see below).
+
+
+
+We next examined the 49 total clusters to ensure that our identified clusters truly represented distinct
+
+cellular classifications, as opposed to over-partitioning. We performed a post-hoc test where we
+
+searched for differentially expressed genes (McDavid et al., 2013) between every pair of clusters
+
+(requiring at least 10 genes, each with an average expression difference greater than 1 natural log value
+
+between clusters with a Bonferroni corrected p<0.01). We iteratively merged cluster pairs that did not
+
+satisfy this criterion, starting with the two most related pairs (lowest number of differentially expressed
+
+genes). This process resulted in 10 merged clusters, leaving 39 remaining.
+
+## Page 21
+
+We then computed average gene expression for each of the 39 remaining clusters, and calculated
+
+Euclidean distances between all pairs, using this data as input for complete-linkage hierarchical
+
+clustering and dendrogram assembly. We then compared each of the 39 clusters to the remaining cells
+
+using a likelihood-ratio test (McDavid et al., 2013) to identify marker genes that were differentially
+
+expressed in the cluster.
+
+
+
+Embedding the Projection Set onto the tSNE Map
+
+We used the computational approach in Shekhar et al. (Shekhar et al., 2014) and Berman et al. (Berman
+
+et al., 2014) to project new cells onto an existing tSNE map.  First, the expression vector of the cell is
+
+reduced to include only the set of highly variable genes, and subsequently centered and scaled along
+
+each gene using the mean and standard deviation of the gene expression in the training set. This scaled
+
+expression vector z (dimensions 1 x 384) is multiplied with the scores matrix of the genes S
+
+(dimensions 384 x 32), to obtain its “loadings” along the significant PCs u (dimensions 1 x 32).  Thus,
+
+                                                           𝑢′  =  𝑧′.𝑆
+
+u (dimensions 1 x 32) denotes the representation of the new cell in the PC subspace identified from the
+
+training set. We note a point of consistency here in that performing the above dot product on a scaled
+
+expression vector of a cell z taken from the training set recovers its correct subspace representation u, as
+
+it ought to be the case.
+Given the PC loadings of the cells in the training set {ui} (i=1,2,…N                    ) and their tSNE coordinates {yi}
+                                                                                     train
+
+(i=1,2,…Ntrain), the task now is to find the tSNE coordinates y’ of the new cell based on its loadings
+
+vector u’. As in the original tSNE framework (van der Maaten and Hinton, 2008), we “locate” the new
+
+cell in the subspace relative to the cells in the training set by computing a set of transition probabilities,
+                                                         exp   (−𝑑(𝑢′,    𝑢𝑖)2⁄2𝜎2     )
+                                      𝑝(𝑢′|𝑢𝑖)     =                                 𝑢′
+                                                       ∑{𝑢𝑖} exp(−𝑑(𝑢′,      𝑢𝑖)2⁄2𝜎𝑢2′)
+
+## Page 22
+
+Here, d(. , .) represents Euclidean distances, and the the bandwidth σu’ is chosen by a simple binary
+search in order to constrain the Shannon entropy associated with 𝑝(𝑢′|𝑢𝑖) to log2(30), where 30
+
+corresponds to the value of the perplexity parameter used in the tSNE embedding of the training set.
+
+Note that σu’ is chosen independently for each cell.
+
+
+
+A corresponding set of transition probabilities in the low dimensional embedding are defined based on
+
+the Student’s t-distribution as,
+
+                                                                         ′   𝑖 2  −1
+                                               ′   𝑖         (1 +   𝑑(𝑦   ,𝑦  )  )
+                                          𝑞(𝑦   |𝑦  ) =   ∑     (1  +  𝑑(𝑦′,  𝑦𝑖)2)−1
+                                                            {𝑦𝑖}
+
+where y’ are the coordinates of the new cell that are unknown. We calculate these by minimizing the
+Kullback-Leibler divergence between 𝑝(𝑢′|𝑢𝑖) and 𝑞(𝑦′|𝑦𝑖),
+
+
+                                         ′                          ′   𝑖       𝑝(𝑢′|𝑢𝑖)
+                                       𝑦   =   𝑎𝑟𝑔𝑚𝑖𝑛 ∑        𝑝(𝑢   |𝑢  )log𝑞(𝑦′|𝑦𝑖)
+                                                            𝑖
+
+This is a non-convex objective function with respect to its arguments, and is minimized using the
+
+Nelder-Mead simplex algorithm, as implemented in the Matlab function fminsearch. This procedure
+
+can be parallelized across all cells in the projection set.
+
+A few notes on the implementation,
+
+     1.   Since this is a post-hoc projection, and 𝑝(𝑢′|𝑢𝑖) is only a relative measure of pairwise
+
+          similarity in that it is always constrained to sum to 1, we wanted to avoid the possibility of new
+
+          cells being embedded on the tSNE map by virtue of their high relative similarity to one or two
+
+          training cells (“short circuiting”). In other words, we chose to project only those cells that were
+
+          drawn from regions of the PC subspace that were well represented in the training set by at least
+
+          a few cells.
+
+          Thus, we retained a cell u’ for projection only if 𝑝(𝑢′|𝑢𝑖)              >  𝑝𝑡ℎ𝑟𝑒𝑠 was true for at least Nmin
+
+          cells in the training set (pthres = 5     × 10−3, Nmin = 10). We calibrated the values for pthres and
+
+## Page 23
+
+          Nmin by testing our projection algorithm on cases where the projection set was known to be
+
+          completely different from the training set to ensure that such cells were largely rejected by this
+
+          constraint. (see Section ‘“Out of sample” projection test’)
+
+     2.   For cells that pass the constraint in pt. 1., the initial value of the tSNE coordinate y’0 is set to,
+
+                                                         𝑦′0  =  ∑𝑝(𝑢′|𝑢𝑖)𝑦𝑖
+
+                                                                  𝑖
+
+          i.e. a weighted average of the tSNE coordinates of the training set with the weights set to the
+
+          pairwise similarity in the PC subspace representation.
+
+     3.   A cell satisfying the condition in 1. is said to be “successfully projected” to a location y’* when
+
+          a minimum of the KL divergence could be found within the maximum number of iterations.
+
+          However since the program is non-convex and is guaranteed to only find local minima, we
+
+          wanted to explore if a better minima could be found. Briefly, we uniformly sampled points
+
+          from a 25 x 25 grid centered on y’* to check for points where the value of the KL-divergence
+
+          was within 5% of its value at y’* or lower. Whenever this condition was satisfied (< 2%) of the
+
+          time, we re-ran the optimization by setting the new point as the initial value.
+
+
+
+“Out of Sample” Projection Test
+
+In order to test our post-hoc projection method, we conducted the following computational experiment
+
+wherein each of the 39 distinct clusters on the tSNE map was synthetically “removed” from the tSNE
+
+map, and then reprojected cell-by-cell on the tSNE map of the remaining clusters using the procedure
+
+outlined above. Only cells from the training set were used in these calculations.
+
+       Assuming our cluster distinctions are correct, in each of these 39 experiments, the cluster that is
+
+being reprojected represents an “out of sample” cell type. Thus successful assignments of these cells
+
+into one of the remaining 38 clusters would be spurious. For each of the 39 clusters that was removed
+
+and reprojected, we classified the cells into three groups based on the result of the projection method:
+
+## Page 24
+
+     (1)  Cells that did not satisfy the condition 1. in the previous section (i.e. did not have a high
+
+          relative similarity to at least Nmin training cells), and therefore “failed” to project.
+
+     (2)  Cells that were successfully assigned a tSNE coordinate y’, but that could not be assigned into
+
+          any of the existing clusters according to the condition below.
+
+     (3)  Cells that were successfully assigned a tSNE coordinate y’, and which were “wrongly
+
+          assigned” to one of the existing clusters. A cell was assigned to a cluster whose centroid was
+
+          closest to y’ if and only if the distance between y’ and the centroid was smaller than the cluster
+
+          radius (the distance of the farthest point from the centroid).
+
+Encouragingly for all of the 39 “out of sample” projection experiments, only a small fraction of cells
+
+were spuriously assigned to one of the clusters, i.e. satisfied (3) above with the parameters pthres =
+5  ×  10−3 and Nmin = 10 (Table S7).  This gave us confidence that our post-hoc embedding of the
+
+projection set would not spuriously assign distinct cell types into one of the existing clusters.
+
+
+
+Downsampling Analyses of Retina Data
+
+
+
+To generate the 500-cell and 2000-cell downsampled tSNE plots shown in Figure 5F, the largest 500 or
+
+2000 cells were sampled from the high-purity replicate (replicate 7), and used as input for PCA and
+
+tSNE.  Two extreme outlier points were removed from the 500-cell tSNE prior to plotting.  To generate
+
+the 9,731-cell downsampled tSNE plot, 10,000 cells were randomly sampled from the full dataset, and
+
+the cells expressing transcripts from more than 900 genes were used in principal components analysis
+
+and tSNE; the remaining (smaller) cells were projected onto the tSNE embedding.
+
+
+
+Immunohistochemistry
+
+## Page 25
+
+Wild-type C57 mice or Mito-P mice, which express CFP in nGnG amacrine and Type 1 bipolar cells
+
+(Kay et al., 2011), were euthanized by intraperitoneal injection of pentobarbital. Eyes were fixed in 4%
+
+PFA in PBS on ice for one hour, followed by dissection and post-fixation of retinas for an additional 30
+
+minutes, then rinsed with PBS. Retinas were frozen and sectioned at 20 μm in a cryostat. Sections were
+
+incubated with primary antibodies (chick anti-GFP [Abcam], rabbit anti-PPP1R17 [Atlas], or goat anti-
+
+VSX2 [Santa Cruz]) overnight at 4°C, and with secondary antibodies (Invitrogen and Jackson
+
+ImmunoResearch) for 2 hours at room temperature. Sections were then mounted using Fluoromount G
+
+(Southern Biotech) and viewed with an Olympus FVB confocal microscope.
+
+
+
+Note on Bead Surface Primers and Custom Sequencing Primers
+
+
+
+During the course of experiments for this paper, we used two batches of beads that had two slightly
+
+different primer sequences (Barcoded Bead SeqA and Barcoded Bead SeqB, Table S6).  Barcoded
+
+Bead SeqA was used in the human-mouse experiments, and in replicates 1-3 of the retina experiment.
+
+Replicates 4-7 were performed with Barcoded Bead SeqB.  To prime read 1 for Drop-Seq libraries
+
+produced using Barcoded Bead SeqA beads, Read1CustSeqA was used; to prime read 2 for Drop-Seq
+
+libraries produced using Barcoded Bead SeqB beads, Read1CustSeqB was used.  ChemGenes plans to
+
+manufacture beads harboring the Barcoded Bead SeqB sequence.  These beads should be used with
+
+Read1CustSeqB.
+
+
+
+
+
+Additional Notes Regarding Drop-Seq Implementation
+
+
+
+Cell and Bead Concentrations
+
+## Page 26
+
+Our experiments have shown that the cell concentration used in Drop-Seq has a strong, linear
+
+relationship to the purity and doublet rates of the resulting libraries (Figures 3A, 3B, and S3B).  Cell
+
+concentration also linearly affects throughput: ~10,000 single-cell libraries can be processed per hour
+
+when cells are used at a final concentration of 100 cells / μL, and ~1,200 can be processed when cells
+
+are used at a final concentration of 12.5 cells / μL.  The trade-off between throughput and purity is
+
+likely to affect users differently, depending on the specific scientific questions being asked.  Currently,
+
+for our standard experiments, we use a final concentration of 50 cells / μL, tolerating a small percentage
+
+of doubles and cell contaminants, to be able to easily and reliably process 10,000 cells over the course
+
+of a couple of hours.  As recommended above, we currently favor loading beads at a concentration of
+
+120 / μL  (final concentration in droplets = 60 / μL), which empirically yields a < 5% bead doublet rate.
+
+
+
+Drop-Seq Start-Up Costs
+
+The main pieces of equipment required to implement Drop-Seq are three syringe pumps (KD Legato
+
+100 pumps, list price ~$2,000 each) a standard inverted microscope (Motic AE31, list price ~$1,900),
+
+and a magnetic stirrer (V&P scientific, #710D2, list price ~$1,200).   A fast camera (used to monitor
+
+droplet generation in real time) is not necessary for the great majority of users (droplet quality can be
+
+monitored by simply placing 3 μL of droplets in a Fuchs-Rosenthal hemocytometer with 17 μL of
+
+droplet generation oil to dilute the droplets into a single plane of focus).
+
+## Page 27
+
+Table S1.  Analysis of edit distance relationships among UMIs, Related to Figure 3
+
+
+         UMI Sampling                                          %  Reduction in UMI counts
+
+                                         Substitution-only collapse              Indel and substitution collapse
+
+         Within a gene                               68.2%                                       76.1%
+
+         Across genes                                19.1%                                       45.7%
+
+Edit distance relationships among UMIs.  For the data in Figure 3G, the sequences of the UMIs for
+
+each ERCC gene detected in each cell barcode were collapsed at an edit distance of 1, including only
+
+substitutions (left column) or with both substitutions and insertions/deletions (right column).  A control
+
+UMI set was prepared for each gene, using an equal number of UMIs sampled randomly across all
+
+genes/cells.  The table shows the percent of the original UMIs that were collapsed for each condition.
+
+## Page 28
+
+Table S5.  Cost Analysis of Drop-Seq, Related to Figure 5
+Reagents                                                          Supplier                Catalog #                  Cost for
+                                                                                                            10,000 cells ($)
+Microfluidics costs (tubing, syringes,                                  N/A                      N/A                    35.00
+droplet generation oil, device fabrication)
+DropSeq lysis buffer (Ficoll, Tris, Sarkosyl,                           N/A                      N/A                      9.35
+EDTA, DTT)
+Barcoded microparticles                                       Chemgenes                          N/A                   137.20
+Maxima H– Reverse Transcriptase                                    Thermo                   EP0753                      59.15
+dNTP mix                                                          Clontech                   639125                       7.78
+RNase inhibitor                                                    Lucigen                  30281-2                       3.80
+Template switch oligo                                                   IDT                      N/A                      7.60
+Perfluorooctanol                                                     Sigma                   370533                     11.90
+Exonuclease I                                                          NEB                  M0293L                        3.84
+KAPA Hifi HotStart ReadyMix                           KAPA BioSystems                       KK2602                     210.00
+Nextera XT DNA sample preparation kit                              Illumina           FC-131-1096                      120.80
+Ampure XP beads                                         Beckman Coulter                     A63882                      37.35
+BioAnalyzer High Sensitivity Chips                                  Agilent              5067-4626                        9.64
+
+
+                                      Total cost:                                                                    $653.41
+                                  Cost per cell:                                                                       $0.065
+
+## Page 29
+
+Table S6.  Oligonucleotide Sequences Used in This Study
+
+
+               synRNA                 rCrCrUrArCrArCrGrArCrGrCrUrCrUrUrCrCrGrArUrCrUrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNrNr
+                                      BrArArArArArArArArArArArArArArArArArArArArArArArA
+
+        Barcoded Bead SeqA            5’ –Bead–Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACGTJJJJJJJJJJJJNNNNNNNN
+                                      TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3’
+        Barcoded Bead SeqB            5’ –Bead–Linker-TTTTTTTAAGCAGTGGTATCAACGCAGAGTACJJJJJJJJJJJJNNNNNNNN
+                                      TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-3’
+      Template_Switch_Oligo           AAGCAGTGGTATCAACGCAGAGTGAATrGrGrG
+              TSO_PCR                 AAGCAGTGGTATCAACGCAGAGT
+           P5-TSO_Hybrid              AATGATACGGCGACCACCGAGATCTACACGCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+           Nextera_N701               CAAGCAGAAGACGGCATACGAGATTCGCCTTAGTCTCGTGGGCTCGG
+           Nextera_N702               CAAGCAGAAGACGGCATACGAGATCTAGTACGGTCTCGTGGGCTCGG
+           Nextera_N703               CAAGCAGAAGACGGCATACGAGATTTCTGCCTGTCTCGTGGGCTCGG
+         Read1CustomSeqA              GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTACGT
+         Read1CustomSeqB              GCCTGTCCGCGGAAGCAGTGGTATCAACGCAGAGTAC
+           P7-TSO_Hybrid              CAAGCAGAAGACGGCATACGAGATCGTGATCGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGT*A*C
+
+              TruSeq_F                AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
+
+          CustSynRNASeq               CGGTCTCGGCGGAAGCAGTGGTATCAACGCAGAGTAC
+
+           UMI_SMARTdT                AAGCAGTGGTATCAACGCAGAGTACNNNNNNNNNTTTTTTTTTTTTTTTTTTTTTTTT
+
+## Page 30
+
+Table S7.  “Out-of-Sample” Projection Test
+
+  Cluster #           # Cells in           # failed to         # Projected          # Wrongly            % Wrongly
+                        Cluster              project                                 Assigned             Assigned
+       1                  153                  153                   0                    0                  0.00
+       2                  271                  271                   0                    0                  0.00
+       3                  201                  201                   0                    0                  0.00
+       4                   46                  46                    0                    0                  0.00
+       5                   63                  62                    1                    0                  0.00
+       6                  173                  156                  17                    9                  5.20
+       7                  277                  272                   5                    5                  1.81
+       8                  115                  115                   0                    0                  0.00
+       9                  275                  275                   0                    0                  0.00
+      10                  155                  153                   2                    2                  1.29
+      11                  165                  162                   3                    3                  1.82
+      12                  175                  175                   0                    0                  0.00
+      13                   46                  40                    6                    5                 10.87
+      14                   89                  89                    0                    0                  0.00
+      15                   52                  44                    8                    6                 11.54
+      16                  179                  179                   0                    0                  0.00
+      17                  284                  284                   0                    0                  0.00
+      18                   64                  63                    1                    1                  1.56
+      19                  108                  107                   1                    0                  0.00
+      20                  206                  206                   0                    0                  0.00
+      21                  154                  154                   0                    0                  0.00
+      22                  180                  180                   0                    0                  0.00
+      23                  183                  182                   1                    1                  0.55
+      24                 3712                 3417                 295                  180                  4.85
+      25                 1095                 1071                  24                   18                  1.64
+      26                 1213                 1212                   1                    0                  0.00
+      27                  323                  318                   5                    4                  1.24
+      28                  339                  330                   9                    7                  2.06
+      29                  332                  324                   8                    6                  1.81
+      30                  447                  426                  21                   18                  4.03
+      31                  346                  340                   6                    3                  0.87
+      32                  235                  233                   2                    2                  0.85
+      33                  453                  450                   3                    3                  0.66
+      34                  784                  784                   0                    0                  0.00
+      35                   27                  27                    0                    0                  0.00
+      36                   43                  43                    0                    0                  0.00
+      37                  145                  139                   6                    5                  3.45
+      38                   30                  30                    0                    0                  0.00
+      39                   17                  17                    0                    0                  0.00
+
+
+For each cluster, the “training” cells were removed from the tSNE plot, and then projected onto the
+tSNE.  The number of cells that successfully projected into the embedding, and the number of cells that
+were inappropriately incorporated into a different cluster were tabulated.
+
+## Page 31
+
+References
+
+Bar-Joseph, Z., Siegfried, Z., Brandeis, M., Brors, B., Lu, Y., Eils, R., Dynlacht, B.D., and Simon, I. (2008).
+Genome-wide transcriptional analysis of the human cell cycle identifies genes differentially regulated in normal
+and cancer cells. Proceedings of the National Academy of Sciences of the United States of America 105, 955-960.
+Barres, B.A., Silverstein, B.E., Corey, D.P., and Chun, L.L. (1988). Immunological, morphological, and
+electrophysiological variation among retinal ganglion cells purified by panning. Neuron 1, 791-803.
+Berman, G.J., Choi, D.M., Bialek, W., and Shaevitz, J.W. (2014). Mapping the stereotyped behaviour of freely
+moving fruit flies. Journal of the Royal Society, Interface / the Royal Society 11.
+Carter-Dawson, L.D., and LaVail, M.M. (1979). Rods and cones in the mouse retina. I. Structural analysis using
+light and electron microscopy. The Journal of comparative neurology 188, 245-262.
+Chung, N.C., and Storey, J.D. (2014). Statistical Significance of Variables Driving Systematic Variation in High-
+Dimensional Data. Bioinformatics.
+Ester, M., Kriegel, H.P., Sander, J., and Xu, X. (1996). A density-based algorithm for discovering clusters in large
+spatial databases with noise. (Menlo Park, Calif.: AAAI Press).
+Islam, S., Zeisel, A., Joost, S., La Manno, G., Zajac, P., Kasper, M., Lonnerberg, P., and Linnarsson, S. (2014).
+Quantitative single-cell RNA-seq with unique molecular identifiers. Nature methods 11, 163-166.
+Kay, J.N., Voinescu, P.E., Chu, M.W., and Sanes, J.R. (2011). Neurod6 expression defines new retinal amacrine
+cell subtypes and regulates their fate. Nature neuroscience 14, 965-972.
+Leek, J.T., and Storey, J.D. (2011). The joint null criterion for multiple hypothesis tests. Applications in Genetics
+and Molecular Biology 10, 1-22.
+Matz, M.V., Alieva, N.O., Chenchik, A., and Lukyanov, S. (2003). Amplification of cDNA ends using PCR
+suppression effect and step-out PCR. Methods in molecular biology 221, 41-49.
+Mazutis, L., Gilbert, J., Ung, W.L., Weitz, D.A., Griffiths, A.D., and Heyman, J.A. (2013). Single-cell analysis
+and sorting using droplet-based microfluidics. Nature protocols 8, 870-891.
+McDavid, A., Finak, G., Chattopadyay, P.K., Dominguez, M., Lamoreaux, L., Ma, S.S., Roederer, M., and
+Gottardo, R. (2013). Data exploration, quality control and testing in single-cell qPCR-based gene expression
+experiments. Bioinformatics 29, 461-467.
+McDonald, J.C., Duffy, D.C., Anderson, J.R., Chiu, D.T., Wu, H., Schueller, O.J., and Whitesides, G.M. (2000).
+Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis 21, 27-40.
+Picelli, S., Bjorklund, A.K., Faridani, O.R., Sagasser, S., Winberg, G., and Sandberg, R. (2013). Smart-seq2 for
+sensitive full-length transcriptome profiling in single cells. Nature methods 10, 1096-1098.
+Shalek, A.K., Satija, R., Adiconis, X., Gertner, R.S., Gaublomme, J.T., Raychowdhury, R., Schwartz, S., Yosef,
+N., Malboeuf, C., Lu, D., et al. (2013). Single-cell transcriptomics reveals bimodality in expression and splicing
+in immune cells. Nature 498, 236-240.
+Shalek, A.K., Satija, R., Shuga, J., Trombetta, J.J., Gennert, D., Lu, D., Chen, P., Gertner, R.S., Gaublomme, J.T.,
+Yosef, N., et al. (2014). Single-cell RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510,
+363-369.
+Shekhar, K., Brodin, P., Davis, M.M., and Chakraborty, A.K. (2014). Automatic Classification of Cellular
+Expression by Nonlinear Stochastic Embedding (ACCENSE). Proceedings of the National Academy of Sciences
+of the United States of America 111, 202-207.
+van der Maaten, L., and Hinton, G. (2008). Visualizing Data using t-SNE. Journal of Machine Learning Research
+9, 2579-2605.
+Whitfield, M.L., Sherlock, G., Saldanha, A.J., Murray, J.I., Ball, C.A., Alexander, K.E., Matese, J.C., Perou,
+C.M., Hurt, M.M., Brown, P.O., et al. (2002). Identification of genes periodically expressed in the human cell
+cycle and their expression in tumors. Molecular biology of the cell 13, 1977-2000.

scrrbs/paper.pymupdf_text.txt

@@ -0,0 +1,679 @@
+# pymupdf text extraction
+source_pdf: paper.pdf
+source_path: /Users/seqmachines/playground/protocols-test/scg-v1-upload/protocols/scrrbs/paper.pdf
+extraction: PyMuPDF: page.get_text("text", sort=True)
+
+## Page 1
+
+          DownloadedDownloaded fromfrom genome.cshlp.orggenome.cshlp.org onon JuneJune 10,10, 20262026 .. PublishedPublished byby ColdCold SpringSpring HarborHarbor LaboratoryLaboratory PressPress
+
+
+
+
+
+Single-cell methylome landscapes of mouse embryonic stem cells
+and early embryos analyzed using reduced representation bisulfite
+sequencing
+Hongshan Guo, Ping Zhu, Xinglong Wu, et al.
+Genome Res. published online October 31, 2013
+Access the most recent version at doi:10.1101/gr.161679.113
+
+          P<P   Published online October 31, 2013 in advance of the print journal.
+
+          Creative   This article is distributed exclusively by Cold Spring Harbor Laboratory Press for the
+     Commons     first six months after the full-issue publication date (see
+         License   http://genome.cshlp.org/site/misc/terms.xhtml). After six months, it is available under                    a Creative Commons License (Attribution-NonCommercial 3.0 Unported), as
+                       described at http://creativecommons.org/licenses/by-nc/3.0/.
+
+  Email Alerting   Receive free email alerts when new articles cite this article - sign up in the box at the
+         Service    top right corner of the article or click here.
+
+
+
+
+
+To subscribe to Genome Research go to:
+https://genome.cshlp.org/subscriptions
+
+
+© 2013 Guo et al.; Published by Cold Spring Harbor Laboratory Press
+
+## Page 2
+
+             Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+Method
+
+Single-cell methylome landscapes of mouse embryonic
+stem cells and early embryos analyzed using reduced
+representation bisulfite sequencing
+
+Hongshan Guo,1,3 Ping Zhu,1,2,3 Xinglong Wu,1 Xianlong Li,1 Lu Wen,1
+and Fuchou Tang1,4
+1Biodynamic Optical Imaging Center, College of Life Sciences, Peking University, Beijing 100871, China; 2Peking-Tsinghua Center
+for Life Sciences, Peking University, Beijing 100871, China
+
+
+    DNA methylation is crucial for a wide variety of biological processes, yet no technique suitable for the methylome
+       analysis of DNA methylation at single-cell resolution is available. Here, we describe a methylome analysis technique that
+      enables single-cell and single-base resolution DNA methylation analysis based on reduced representation bisulfite se-
+      quencing (scRRBS). The technique is highly sensitive and can detect the methylation status of up to 1.5 million CpG sites
+      within the genome of an individual mouse embryonic stem cell (mESC). Moreover, we show that the technique can detect
+      the methylation status of individual CpG sites in a haploid sperm cell in a digitized manner as either unmethylated or fully
+      methylated. Furthermore, we show that the demethylation dynamics of maternal and paternal genomes after fertilization
+      can be traced within the individual pronuclei of mouse zygotes. The demethylation process of the genic regions is faster
+      than that of the intergenic regions in both male and female pronuclei. Our method paves the way for the exploration of
+      the dynamic methylome landscapes of individual cells at single-base resolution during physiological processes such as
+     embryonic development, or during pathological processes such as tumorigenesis.
+
+      [Supplemental material is available for this article.]
+
+Gene transcription is crucial for a cell to maintain its identity and    methylomes of mammalian cells (Meissner et al. 2005). RRBS is
+physiological function and is regulated within individual cells.    based on the lack of even distribution for CpG sites within the
+Epigenetic status is important in transcriptional regulation and is    mammalian genome; these sites tend to cluster together as CpG
+potentially heterogeneous even within a relatively homogeneous     islands (CGIs) that are usually located near the promoter regions of
+cell type due to the different cell subpopulations present (Jaenisch    annotated genes (Deaton and Bird 2011). Thus, after cutting the
+and Bird 2003; Toyooka et al. 2008). This is especially prominent in    genome into short fragments via a restriction enzyme that recog-
+tumors in which both the genomes and epigenomes of the in-     nizes CpG and its flanking sequences, a majority of the CGIs will
+dividual cells are heterogeneous (Rodriguez-Paredes and Esteller    be recovered and sequenced with high coverage even with rela-
+2011; Marusyk et al. 2012). Moreover, it is very difficult to obtain     tively low numbers of total sequencing reads. RRBS has been shown
+large numbers of cells for epigenome analysis in some circum-     to be effective for as few as 60 mouse early embryonic cells (Chan
+stances, such as for mammalian early embryos (Smallwood et al.     et al. 2012; Smallwood and Kelsey 2012) and has led to significant
+2011; Tang et al. 2011a; Smith et al. 2012). It is highly desirable to     findings regarding global demethylation and remethylation pro-
+develop a single-cell epigenome analysis technique, ideally one     cesses during the early cleavage and post-implantation stages of
+that provides single-base resolution. As one of the most important    mouse embryonic development, respectively (Smith et al. 2012).
+epigenetic modifications, DNA methylation is critical for a wide     Recently, a method for the epigenetic analysis of histone modifi-
+variety of biological processes, including the regulation of genomic     cations for an individual locus at single-cell resolution has been
+imprinting and X-chromosome inactivation, as well as the re-    developed (Gomez et al. 2013). However, single-cell epigenome
+pression of transposable elements within the genome (Bird 2002;     analysis at whole-genome scale has never been achieved.
+Lister et al. 2009; Hackett et al. 2012). DNA is methylated at the      We report the development of a single-cell methylome ana-
+carbon atom occupying the fifth position of the cytosine ring      lysis technique based on RRBS (scRRBS) and demonstrate its ef-
+(5mC) and is catalyzed by the DNA cytosine methyltransferases,     fective use for mouse embryonic stem cells (mESCs), sperm, and
+Dnmt1, Dnmt3a, and Dnmt3b (Reik 2007). DNA methylation is     oocytes, as well as for male and female pronuclei of the zygotes. We
+functionally important for mammalian development because    were able to recover 0.5 to 1.5 million CpG sites from a single
+both Dnmt1 and Dnmt3b knockout mice are embryonic lethal,    mESC, and the methylation levels for all analyzed genomic regions
+whereas Dnmt3a mutant mice die within 1 mo after birth (Okano    were comparable to those obtained from bulk mESCs (Table 1;
+et al. 1999; Li 2002). The reduced representation bisulfite se-    Supplemental Table 1). Furthermore, we show for the first time
+quencing (RRBS) technique has been developed to dissect the     that the methylome of the first polar body is comparable with that
+
+
+
+3These authors contributed equally to this work.                       2013 Guo et al.  This article is distributed exclusively by Cold Spring Harbor
+4Corresponding author                                                     Laboratory Press for the first six months after the full-issue publication date (see
+E-mail tangfuchou@pku.edu.cn                                              http://genome.cshlp.org/site/misc/terms.xhtml). After six months, it is avail-
+Article published online before print. Article, supplemental material, and publi-      able under a Creative Commons License (Attribution-NonCommercial 3.0
+cation date are at http://www.genome.org/cgi/doi/10.1101/gr.161679.113.        Unported), as described at http://creativecommons.org/licenses/by-nc/3.0/.
+
+
+23:000–000 Published by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/13; www.genome.org             Genome Research  1
+                                                                                              www.genome.org
+
+## Page 3
+
+              Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+Guo et al.
+
+
+
+Table 1.  Summary of the unique covered CpG sites and their mean coverage depths at 13, 53, and 103 in each single mESC cell and in bulk
+mESCs
+
+                    Unique       Mean         Unique       Mean         Unique        Mean               Bisulfite
+Sample           CpGs (13)    coverage (13)   CpGs (53)    coverage (53)   CpGs (103)    coverage (103)    conversion rate
+
+
+ESC single-cell 1       1,309,191         63           776,187        105           633,316         128            98.43%
+ESC single-cell 2        955,619         21           428,819         44           318,552          57            99.57%
+ESC single-cell 3       1,056,351         31           518,658         62           400,237          78            99.15%
+ESC single-cell 4       1,535,234         64           940,853        104           775,101         124            99.94%
+ESC single-cell 5       1,269,763         13           607,502         25           410,319          33            98.89%
+ESC single-cell 6        970,525         38           462,859         79           355,307         100            97.74%
+ESC single-cell 7        496,715         31           231,857         65           171,483          86            99.49%
+ESC single-cell 8        573,049         23           282,939         45           208,351          58            99.48%
+ESC pooled 5-cell      1,853,963         39          1,160,049         61           930,025          74            97.98%
+ESC pooled 10-cell    2,381,797         40          1,445,754         64          1,195,001          76            99.73%
+ESC pooled 20-cell    2,592,919         45          1,801,643         64          1,498,989          76            99.45%
+Bulk mESCs           2,411,401         17          1,476,724         26          1,168,476          32           NA
+
+The right-hand column shows the bisulfite conversion rate of each sample. (NA) Not applicable.
+
+
+of the metaphase II oocyte within the same gamete. Finally, we     1.53 million CpG sites were recovered by merging eight individual
+used our method to prove that the demethylation process of the    mESCs) (Fig. 2A). Similarly, if we merged the scRRBS data of five
+male pronucleus occurs more quickly than that of the female     single sperm cells together, we were able to capture 0.73 million
+pronucleus in the same zygote.                         CpG sites (Supplemental Fig. 2; also see below). Moreover, our
+                                                     method is also applicable to small numbers of cells rather than
+                                                                     individual cells. We showed that using 20 mESCs as the starting
+Results                                                               material, our method could detect 63% (1.52 million) of the CpG
+                                                                                  sites that are recovered using RRBS with bulk mESCs (Fig. 2A;
+Characterization of the single-cell RRBS methylome analysis
+                                                             Supplemental Table 3).
+technique                                               We then determined the accuracy of our method and com-
+To improve the suitability of the RRBS method for single-cell    pared the data obtained using individual mESCs with that
+analysis, we reasoned that one of the primary obstacles to success    obtained using bulk mESCs, and found that the correlation co-
+was the massive loss of DNA during multiple purification steps.     efficient was reasonably high (R = 0.77 on average) when com-
+Thus, we thoroughly modified the original protocol (Smith et al.     paring the CpG sites recovered using both methods (Fig. 2B; Sup-
+2009; Gu et al. 2011a) and integrated all of the experimental pro-    plemental Fig. 3). Moreover, when we merged the single-cell RRBS
+cesses in a single-tube reaction without including any purification     data from all eight individual mESCs, the correlation coefficient
+steps prior to the bisulfite conversion process. That is, all of the    between the merged single-cell data and the data from the bulk
+following steps were completed within the same reaction tube: the    mESCs obtained was 0.90 (Supplemental Fig. 1). Furthermore,
+lysis of an individual cell; the release of naked, double-stranded    unsupervised hierarchical clustering analysis showed that the
+genomic DNA; adding a spike-in of lambda DNA; digestion of the    methylomes of individual mESCs clearly clustered together with
+genomic DNA using a restriction enzyme; the end-repair and dA-     those of bulk mESCs but remained separate from those of oocytes,
+tailing of the DNA fragments; ligation of the adaptors to the DNA    sperm, and male and female pronuclei (Fig. 2B; also see below).
+fragments; and the bisulfite conversion of the ligated DNA. After     Figure 2C and Supplemental Figure 3 show several loci that are
+this procedure, the converted DNA was purified using Zymo spin     representative of the methylation status of the individual and bulk
+columns with 10 ng tRNA as a carrier. The purified DNA was then    mESCs. Furthermore, the methylation levels measured for specific
+enriched via two rounds of PCR amplification and subjected to    genomic regions of individual mESCs are similar to those of
+deep sequencing (for details, see Methods) (Fig. 1).                 bulk mESCs (Fig. 2D; Supplemental Fig. 4). These results in-
+    Whether the conversion rate for the trace amounts of DNA     dicate that our method can be used to accurately analyze the
+obtained from a single cell after being treated with bisulfite is     global DNA methylation status within individual cells. We then
+comparable to that of bulk DNA had not previously been de-    sought to determine whether the method is robust by com-
+termined. By spiking trace amounts of unmethylated lambda DNA    paring the methylome of different individual mESCs; the resulting
+into the single-cell samples, we determined that the conversion     correlation coefficient was reasonably high among individual
+rate resulting from bisulfite treatment is 99.2% on average (ranging    mESCs (R = 0.67 on average), verifying that our method was clearly
+from 97.7% to 99.9%), indicating that the DNA from each single     reproducible (Fig. 2B).
+cell was converted highly efficiently under our bisulfite treatment      We then sought to determine whether the scRRBS method
+condition (Table 1; Supplemental Table 1).                          enables us to obtain the absolute methylation status of a CpG site.
+   We then determined how many CpG sites could be recovered    In theory, our method should only detect a CpG locus as either
+using our scRRBS approach. We analyzed eight individual mESCs     fully methylated (100%) or unmethylated (0%) but not as an in-
+and determined the presence of 496,715 to 1,535,234 CpG sites in     termediate methylation value (e.g., 30% methylation) in a haploid
+each cell (Table 1). That is, our approach recovered on average 1.02      cell such as sperm, as can be detected in bulk analysis due to the
+million (40%) of the total 2.5 million CpG sites that could be     heterogeneity within the cell population. To investigate this as-
+detected using RRBS with bulk cells (Supplemental Table 2; Smith     pect, we applied our method to single sperm cells. We found that
+et al. 2012). As expected, when we merged the RRBS data for in-   88%–94% of the CpG sites detected using our method within an
+dividual cells together, additional CpG sites were recovered (e.g.,     individual sperm  cell are either  fully methylated (100%) or
+
+
+2  Genome Research
+      www.genome.org
+
+## Page 4
+
+             Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+                                                                  Single-cell methylome assay of mouse early embryos
+
+
+
+
+
+Figure 1.  A schematic of the single-cell RRBS (scRRBS) technique. Note the completion of all of the following steps within the same reaction tube: lysis
+of an individual cell, release of the naked double-stranded genomic DNA, spiking with lambda DNA, digestion of the genomic DNA using a restriction
+enzyme, end-repair and dA-tailing of the DNA fragments, ligation of the adaptors to the DNA fragments, and bisulfite conversion of the ligated DNA.
+
+
+unmethylated (0%) (Fig. 3A; Supplemental Fig. 5), indicating that     the methylated locus can be digested using MspI (methylation-in-
+our method is accurate and digitized. By applying the same strat-     sensitive restriction enzyme) but not by HpaII (methylation-sen-
+egy, we found that 84%–90% of the CpG sites are either fully      sitive restriction enzyme), whereas the unmethylated locus can be
+methylated or unmethylated in single mESCs (Supplemental     digested by both enzymes (Fig. 3C). This indicates that the meth-
+Fig. 6). Figure 3B and Supplemental Figure 5 show several loci that     ylation status obtained using scRRBS is accurate and can be verified
+are representative of the methylation status in individual sperm    using an independent approach within individual cells.
+cells and bulk sperm. To verify this result using an independent
+approach, we analyzed one fully methylated locus and one un-
+                                                          Applying the analysis to male and female pronucleimethylated locus in single sperm cells and validated their meth-
+ylation status via methylation-sensitive restriction digestion cou-     After fertilization, the maternal and paternal genomes of the
+pled with nested PCR within individual sperm cells. We found that     zygote underwent different types of global demethylation; the
+
+
+                                                                 Genome Research  3
+                                                                                              www.genome.org
+
+## Page 5
+
+              Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+Guo et al.
+
+
+
+
+
+Figure 2.  The sensitivity and reproducibility of the single-cell RRBS technique. (A) The number and proportion of CpG sites detected in the merged
+single mESC RRBS data set overlapped with those from the RRBS of the bulk mESCs. (B) Pearson correlation heatmap among the methylomes of all RRBS
+samples of single cells, pooled cells, or bulk cells. The color key from green to red indicates low to high correlation, respectively. (C ) DNA methylation map
+of the CpG sites at a representative locus in the RRBS data from eight single cells and bulk mESCs. The upward blue bars and downward red bars indicate
+methylated CpGs and unmethylated CpGs, respectively. (D) The methylation levels of different genomic regions of single mESCs; pooled mESCs of five
+cells, 10 cells, and 20 cells; and bulk mESCs.
+
+
+
+former was  passively demethylated  during DNA  replication,    were never analyzed separately from those of the metaphase II oo-
+whereas the latter was primarily actively demethylated by TET3     cytes. We compared the first polar body and the metaphase II oocyte
+oxidation of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine    within the same female gamete and found that their methylomes
+(5hmC) (Mayer et al. 2000; Okada et al. 2010; Gu et al. 2011b;    were very similar and clustered together in the unsupervised hier-
+Wossidlo et al. 2011; Hackett et al. 2013). However, this biological     archical clustering analysis (Fig. 2B). Furthermore, when we ana-
+process has not been investigated at the single-cell level on a ge-     lyzed DNA methylation for different genomic regions, we found that
+nome-wide scale. We addressed this issue by applying our scRRBS     the methylation levels of specific genomic regions, such as the in-
+method to individual female and male pronuclei isolated from the     tragenic or intergenic regions measured in the first polar bodies, were
+same zygotes. First, we sought to determine whether the methylomes    comparable to those measured in metaphase II oocytes (Fig. 4A).
+of the first polar bodies were similar to those of the metaphase II      We then chose zygotes at different pronucleus stages and
+oocytes; in previous studies, the methylomes of the polar bodies    performed single-cell RRBS analyses separately for both the male
+
+
+4  Genome Research
+      www.genome.org
+
+## Page 6
+
+             Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+                                                                  Single-cell methylome assay of mouse early embryos
+
+
+
+
+
+Figure 3.  The methylation status of single sperm cells. (A) The proportion of fully methylated ($90% methylated with read depths greater than or equal
+to three) and unmethylated (#10% methylated with read depths greater than or equal to three) CpG sites within the total CpG sites covered in the scRRBS
+of an individual sperm cell. (B) The methylation status of a representative locus on chromosome 1 showing that most of the detected CpG sites were either
+methylated or unmethylated. Filled black circles represent methylated CpG sites, whereas open circles represent unmethylated CpG sites. Gaps in the
+methylation profiles represent CpG sites that were not recovered in the single-cell RRBS data. The filled brown circles represent all of the CpG sites in the
+corresponding region of the genome. (C ) Agarose gel analysis of the methylation-sensitive restriction digestion coupled with nested PCR in single sperm
+cells. (Top) A methylated locus. (Bottom) An unmethylated locus. The first five lanes (excluding the marker lane) indicate five individual single sperm cells
+digested with MspI, MspI, HpaII, HpaII, and no enzyme, respectively. The next two lanes indicate 1 ng of bulk sperm genomic DNA treated with MspI or
+HpaII, respectively, as positive controls. A weaker band (468 bp) at the upward side of the strong band (320 bp) in the fourth lane (excluding the marker
+lane) of the top panel is the amplification product of the first-round PCR.
+
+
+
+and female pronuclei. The female pronuclei were derived from    Santos et al. 2002; Farthing et al. 2008; Iqbal et al. 2011; Inoue et al.
+metaphase II oocytes, whereas the male pronuclei were derived     2012). This is also consistent with the fact that both the male and
+from sperm. Unsupervised hierarchical clustering analysis showed    female pronuclei passively demethylated their genomes by di-
+that all of the female pronuclei clustered properly with metaphase     lution due to DNA replication during the pronucleus stages,
+II oocytes, whereas all of the male pronuclei clustered together    whereas the male pronuclei also underwent active global demeth-
+with sperm (Fig. 2B). This shows that our scRRBS analysis can be     ylation via TET3 at the same time (Ferreira and Carmo-Fonseca
+used to clearly determine the methylome of individual female and    1997; Gu et al. 2011b).
+male pronuclei. Furthermore, we found that during zygotic de-           Finally, we sought to determine whether the demethylation
+velopment, the methylation levels of the pronuclei decreased sig-     process was synchronized in the genome or whether some geno-
+nificantly as they became closer to each other (Fig. 4B,C). More    mic regions would demethylate faster than other regions. We
+importantly, the demethylation of the male pronuclei was more    found that during the development of the zygotes through the
+dramatic than that of the female pronuclei, which is consistent    pronucleus stages, the methylation levels of the genic regions de-
+with previous immunostaining results and bisulfite sequencing     creased faster than those of the intergenic regions in both the fe-
+results of several individual loci that showed faster demethylation    male (from 20% to 15%) and male pronuclei (from 21% to 10%)
+of male pronuclei than female pronuclei (Oswald et al. 2000;      (Fig. 5A; Supplemental Table 4). This is consistent with the possi-
+
+
+                                                                 Genome Research  5
+                                                                                              www.genome.org
+
+## Page 7
+
+              Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+Guo et al.
+
+
+
+
+
+Figure 4.  Global demethylation in male and female pronuclei during pronucleus stages within individual zygotes. (A) The DNA methylation levels of
+different genomic regions of metaphase II oocytes and the first polar bodies within the same gametes. (B) Hoechst 33342 staining of pronuclei in the
+zygotes, indicating the distance between each pair of male and female pronuclei in individual zygotes. From zygote 1 to zygote 5, the distance between
+the male and female pronuclei gradually decreases. (C ) Global methylation levels in male and female pronuclei within the same zygotes. Note that the
+methylation levels decrease significantly in both male (red line) and female (blue line) pronuclei.
+
+
+
+bility that the genic regions were replicated earlier during the     Recently, single-cell genome sequencing techniques have also
+replication of the maternal and paternal genomes and were    been developed that enable researchers to analyze the heteroge-
+therefore passively demethylated earlier than the intergenic re-     neity of single-cell genomes (Kalisky and Quake 2011; Navin et al.
+gions (Ferreira and Carmo-Fonseca 1997; Iqbal et al. 2011). Several    2011; Zong et al. 2012). Here, we report the development of a sin-
+representative loci indicating significant demethylation of the      gle-cell methylome analysis technique based on RRBS that enables
+male pronuclei are shown in Figure 5B and Supplemental Figure 7.    us to dissect the complexity of the methylomes within an in-
+For the repeat elements in the male pronucleus, the methylation     dividual cell. To our knowledge, this represents the development of
+levels of the short interspersed nuclear elements (SINEs) decreased    the first single-cell epigenome analysis technique. Our single-cell
+more quickly, from 60% in sperm to 35% in the late pronucleus    RRBS technique can be directly applied either to a single cell or to
+stage; in contrast, the methylation levels of the long interspersed    a pool of a small number of cells. Both strategies have advantages
+nuclear elements (LINEs) decreased more slowly, from 81% to 67%    and disadvantages. The RRBS of a single cell is a digitized method
+(Fig. 5C; Supplemental Table 4). This indicates that during the     that can exclude the effect of the heterogeneity of a population of
+demethylation process of repeat elements in the paternal genome,      cells. However, the coverage of this method is relatively low, and
+SINEs were demethylated faster than LINEs and long terminal    the cost for sequencing all of the libraries of these single cells is
+repeats (LTRs) (Xu et al. 2011). During the pronucleus stages, high-     relatively high. On the other hand, the RRBS of a pool of a few
+density CpG promoters (HCP), intermediate-density CpG pro-      cells is not a digitized method, and the unknown heterogeneity
+moters (ICP), and low-density CpG promoters (LCP) do not     of a population of cells will obscure the interpretation of the
+experience significant demethylation in either male or female    methylation status of these loci within individual cells. However,
+pronuclei (Supplemental Fig. 8). The results clearly show that our    the RRBS of a pool of a few cells has relatively high coverage, and
+scRRBS method can be used to trace the methylome of mammalian    the sequencing cost is relatively low because only one or two se-
+cells with single-cell and single-base resolution.                    quencing libraries in the population need to be sequenced.
+                                                          Our scRRBS method has several advantages. First, we spiked
+Discussion                                                            trace amounts of unmethylated lambda DNA into the samples to
+                                                                     accurately measure the bisulfite conversion rate for each single-cell
+Single-cell genomics is crucial to understanding the gene regula-    sample. This enabled us to maintain a very low percentage (<0.8%)
+tion networks within individual cells, which are the fundamental     of false-positive determinations of DNA methylation due to the
+biological units of organisms. We and others developed single-    non-conversion of unmethylated cytosines that were perceived
+cell RNA-seq transcriptome analysis techniques several years ago     falsely as methylated cytosines. Second, our method is highly
+that enable gene expression dynamics to be analyzed within an     sensitive and can detect 40% of the CpG sites from a single cell
+individual cell (Kurimoto et al. 2006; Tang et al. 2009, 2011b).    compared with RRBS using bulk cells (Supplemental Table 2).
+
+
+6  Genome Research
+      www.genome.org
+
+## Page 8
+
+             Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+                                                                  Single-cell methylome assay of mouse early embryos
+
+
+
+
+
+Figure 5.  The demethylation patterns in various genomic regions in male and female pronuclei during zygotic development. (A) DNA methylation
+dynamics of male and female pronuclei in intragenic and intergenic regions. The methylation levels in the male (red line) and female (blue line)
+pronuclei decreased, whereas the demethylation in the male pronuclei was more dramatic than that in the female pronuclei. (B) The methylation
+profile of a representative Meox2 locus on chromosome 12 in the male pronuclei. The upward blue bars in the left panel represent fully methylated CpG
+sites, whereas the downward red bars represent unmethylated CpG sites. The green bar in the panel on the right shows the average methylation levels
+of the CpG sites in this region. (C ) DNA methylation dynamics of male and female pronuclei in repeat regions. The left, middle, and right panels display
+the methylation levels of male and female pronuclei in the SINE, LINE, and LTR regions, respectively. Red and blue lines indicate male and female
+pronuclei, respectively.
+
+
+
+
+Third, our method has intrinsic controls with a cytosine called as    sperm cells, we determined that 91.6% of all detected CpG sites
+fully methylated, unmethylated, or undetected. This is crucial for    were either unmethylated or fully methylated (Fig. 3A). For the
+single-cell epigenome analysis; if we can only call a cytosine as     diploid mESCs, the unmethylated or fully methylated CpG sites
+methylated but cannot discriminate unmethylated cytosines from    accounted for 86.9% of all detected loci (Supplemental Figure 6).
+undetected cytosines due to sample losses, a very high rate of false   One possible explanation for this finding is that ;4.7% of the CpG
+negatives will appear in the assay. This is especially relevant for      sites within a single mESC are likely to be present as one unmeth-
+techniques based on ChIP-seq. For accurate measurements to be     ylated allele together with one fully methylated allele. This phe-
+obtained using a single-cell ChIP-seq technique, the method must   nomenon deserves further study.
+be able to discriminate between the histone marks of the un-        Our method has limitations. Because it is based on the RRBS
+modified status and the undetected status due to sample losses.     technique, it can detect 10% of the CpG sites in the entire genome
+Fourth, our method is flexible and permits both single-cell     at most, leaving 90% of the CpG sites as intractable (Gu et al. 2010,
+methylome analysis and the pooling of a small amount of cells to     2011a). By combining all of the steps prior to PCR amplification
+obtain measurements for the population as a whole. Fifth, our     into a one-tube reaction, we maximally reduce the sample losses
+method is based on the RRBS technique, which can strongly enrich     arising from the use of multiple purification steps. However, the
+CpG-dense sites in the genome; thus, a relatively low number of    dramatic DNA degradation that occurs during bisulfite conversion
+sequence reads is required to detect these target CpG sites at high     (another major potential hurdle to single-cell RRBS) remains un-
+coverage (Gu et al. 2010, 2011a). This makes it feasible to sequence     resolved. In the future, additional strategies are needed to over-
+the methylomes of hundreds or even thousands of single-cell    come this problem and to further improve the coverage of single-
+samples. Sixth, our method performs a digital count of the meth-      cell methylome analysis techniques. Moreover, when 20 mESCs
+ylation status of the CpG sites within a single cell. For the haploid    were pooled together, we recovered only 63% of the CpG sites
+
+
+                                                                 Genome Research  7
+                                                                                              www.genome.org
+
+## Page 9
+
+              Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+Guo et al.
+
+
+
+compared with those recovered using RRBS with bulk mESCs     this enzyme specifically recognizes and cuts unique DNA se-
+(Fig. 2A). It is possible that some CpG sites are more difficult for    quences (C^CGG). The digested DNA was then filled-in and tailed
+our scRRBS technique to detect. It is also possible that bias existed    with an extra A to the 39 blunted ends in a 20 mL reaction in the
+from cell to cell when a small number of cells were pooled to-     presence of 5 units of Klenow fragment (exo-; Fermentas), sup-
+gether for RRBS using our method. Second, we cannot discrimi-    plemented with 1 mM dATP (New England Biolabs), 0.1 mM dGTP
+nate between 5mC and 5hmC using bisulfite sequencing alone    (New England Biolabs), and 0.1 mM dCTP (New England Biolabs).
+                                                                 Illumina standard premethylated indexed adaptors were then li-(Wossidlo et al. 2011; Bock 2012), although in the majority of
+                                                                 gated with the dA-tailed DNA fragments in the presence of 30 unitssomatic cell types, the frequency of 5hmC modifications is sig-
+                                                                      of highly concentrated T4 DNA ligase (Fermentas) in a 25 mL re-
+nificantly lower than that of 5mC (Wu and Zhang 2011). In the
+                                                                       action. Bisulfite conversion was performed in a 150 mL reaction
+future, it may be possible to combine our single-cell RRBS method
+                                                                 using the MethyCode bisulfite conversion kit (Invitrogen) fol-
+with the TAB-seq or oxBS-seq technique to develop a method that
+                                                             lowing the manufacturer’s standard protocol: First, we denatured
+can detect single-cell hydroxymethylomes (Booth et al. 2012; Yu
+                                                                the double-stranded DNA for 10 min at 98°C; then we incubated
+et al. 2012). Third, the RRBS technique provides relatively poor                                                                the reaction for 2.5 h at 64°C to ensure full bisulfite conversion. All
+coverage for imprinting loci in general. Thus, our scRRBS method                                                                      of these steps were performed in a PCR thermocycler. After this
+cannot clearly identify the imprinting status of these loci within                                                                           step, the converted DNA was subjected to on-column desulfona-
+an individual cell. To our knowledge, no DNA methylation assay     tion and purified using Zymo-Spin columns (Zymo) with 10 ng
+for even just an individual imprinting locus within an individual    tRNA (Roche) as a carrier; the DNA was finally eluted in 30 mL of
+cell is currently available.                                           elution buffer. The purified DNA was then subjected to two rounds
+                                                                      of amplification in 50 mL reactions using 1 unit of uracil stalling-
+Methods                                                                free PfuTurbo Cx polymerase (Stratagene) in the first round of PCR
+                                                       and 1 unit of Phusion high-fidelity DNA polymerase (New England
+                                                                       Biolabs) in the second round of PCR. The conditions for the firstIsolation of zygotes, oocytes, and sperm
+                                                         round of PCR were as follows: 2 min at 95°C, followed by 25 cycles
+Four- to five-week-old female C57BL/6N-strain mice were injected     of 20 sec at 95°C, 30 sec at 60°C, and 1 min at 72°C. The conditions
+initially with 5 IU PMSG (Sigma), followed by 5 IU hCG (Sigma)     for the second round of PCR were as follows: 2 min at 98°C, fol-
+45 h later to super-ovulate the mature oocytes. These super-ovu-    lowed by 22 cycles of 10 sec at 98°C, 30 sec at 60°C, and 1 min at
+lated mice were either euthanized to collect oocytes or mated with    72°C. After PCR enrichment, DNA fragments between 200 and 500
+129S2/Sv male mice to obtain male and female pronuclei from the    bp were size-selected and recovered after resolving on a 12% native
+zygotes. The metaphase II oocytes were collected from the oviduct    polyacrylamide TBE gel. The final libraries were assessed using
+ampulla, and the naked oocytes and polar bodies were obtained via    Fragment Analyzer (Advanced Analytical Technologies) to check
+treatment with acidic Tyrode’s solution (Sigma) to remove the zona     size distributions (Supplemental Fig. 10) and quantified using
+pellucida. The spermatozoa were obtained from the caudal epi-    a standard curve-based qPCR assay (Agilent). To exclude the
+didymides of adult 129S2/Sv male mice. Only spermatozoa that     possibility of significant contamination of the scRRBS, we per-
+swam up in HTF medium (Quinn’s Advantage) with vigorous    formed negative controls by omitting the single cell during the
+motility were collected for further RRBS library constructions. All     cell-picking step. That is, we only transferred the carryover buffer
+isolated cells were washed several times in 0.1% PBS-BSA solution     into the lysis buffer and performed all of the following steps in the
+to avoid any possible somatic contaminants. Pronuclei at different    same way as for the scRRBS samples. Three samples of negative
+stages were collected from the zygotes precisely via a daily vaginal     controls were used and prepared as sequencing libraries, and
+plug check. The zygotes were obtained by treating with hyal-     these were sequenced at a depth comparable to that of the single-
+uronidase (Sigma) to remove any attached cumulus cells. After      cell samples. We mapped the data to the mouse genome using
+staining with 5 mg/mL Hoechst 33342 (Invitrogen) for 10 min,    the same parameters as the single-cell samples and found that
+visible pronuclei were isolated by applying a Piezo Micromanipu-    the contamination in these negative control samples was negli-
+lator (PrimeTech)–assisted biopsy (Supplemental Fig. 9). Male and     gible (Supplemental Fig. 11). This rigorously proved that our
+female pronuclei were distinguished based on their relative dis-    scRRBS method  is generally free of contamination. The final
+tances to the polar bodies.                                          quality-ensured libraries were used for pair-ended deep sequenc-
+                                                                 ing on an Illumina HiSeq2000 Sequencer, and all clusters that
+                                                                passed the filter were converted into FASTQ files using a standard
+Culture of mESCs
+                                                                Illumina pipeline. When starting with bulk cells, we constructed
+The mESCs were maintained without feeders on gelatinized dishes     bulk-cell RRBS libraries according to previously published pro-
+in the presence of 1000 units/mL leukemia inhibitory factor (LIF;     tocols (Gu et al. 2011a).
+Millipore) in Dulbecco’s modified eagle’s medium (DMEM/F-12;
+GIBCO) supplemented with 20% fetal calf serum (FCS; GIBCO) for
+                                                                            Single-cell methylation-sensitive digestion coupledroutine passage without any modifications, as previously described
+(Bao et al. 2009).                                                with nested PCR
+                                                            Spermatozoa were isolated from the caudal epididymides of adult
+                                                             129S2/Sv male mice, and single sperm cells were picked and lysed
+Construction of single-cell RRBS sequencing libraries                                                                   in 5 mL of lysis buffer (the same as for scRRBS) and then treated
+Single cells were transferred into 5 mL of lysis buffer (20 mM Tris-    with 9 units of MspI (Fermentas) or 9 units of HpaII (Fermentas) in
+EDTA [pH 8.0], 20 mM KCl, and 0.3% Triton X-100) using a mouth    an 18 mL reaction volume for 3 h at 37°C, respectively. The samples
+pipette, 1 mg/mL protease (Qiagen) was added, and 60 fg unmeth-    were then subjected to nested PCR without purification. The
+ylated lambda DNA (Fermentas) was then spiked in. The cells were     conditions for the first round of PCR were as follows: 5 min at 94°C
+then lysed for 3 h at 50°C and then heat-inactivated for 30 min     followed by 30 cycles of 30 sec at 94°C, 30 sec at 55°C, and 45 sec at
+at 75°C. The released naked DNA was then incubated with    72°C in 100 mL reaction volumes. One microliter of the first-round
+9 units MspI (Fermentas) in an 18 mL reaction for 3 h at 37°C;   PCR product was then used as a template for a 20 mL second-round
+
+
+8  Genome Research
+      www.genome.org
+
+## Page 10
+
+             Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+                                                                  Single-cell methylome assay of mouse early embryos
+
+
+
+PCR. The conditions for the second round of PCR were as follows:     (0%), respectively. CpG sites with methylation levels ranging from
+5 min at 94°C followed by 30 cycles of 30 sec at 94°C, 30 sec at   10% to 90% were discarded in the subsequent analysis. The
+60°C, and 45 sec at 72°C. The final PCR products were visualized    methylation level of the sampled single cell was further described
+on a 1.5% agarose gel (Fig. 3C). For a positive control, we used 1 ng     as the proportion of fully methylated CpG sites to the total CpG
+of bulk sperm genomic DNA as the template for one round of PCR      sites that we covered. Regarding the other RRBS data sets (using
+amplification (the PCR conditions were the same as those used for    more than two cells or bulk cells as starting materials; for example,
+the second round of the single-cell sample PCR). The primers used    the pooled groups of five, 10, and 20 ESCs or bulk ESCs), the fol-
+for the nested PCR are listed in Supplemental Table 5.              lowing cutoffs were applied: CpG sites with less than 10-fold
+                                                                 coverage were discarded, and the remaining informative CpG sites
+                                                            were retained for further analysis.
+Data processing
+
+First, the raw pair-end FASTQ reads were trimmed to remove the
+                                                                  Calculation of the methylation levels of the annotated
+adapter sequences and low-quality bases. The remaining truncated
+                                                         genomic regionsreads were then aligned to the mm9 mouse reference genome
+(downloaded from the UCSC Genome Browser) using the Bismark    The methylation level of each annotated genomic region in each
+tool (http://www.bioinformatics.bbsrc.ac.uk/projects/bismark/)    sample was measured as the sum of the methylation level of every
+(Krueger and Andrews 2011) with the default parameters and ap-   CpG site divided by the total number of the CpG sites that we
+plying a customized pairwise alignment Perl script (Supplemental    covered in the given region. CpG sites with less than 33 coverage
+Table 6). The 48,502-bp lambda DNA genome was built as an extra     in the single-cell RRBS data set or less than 103 coverage in the
+reference to calculate the bisulfite conversion rate. When we esti-    pooled groups of five, 10, and 20 ESCs or bulk ESCs RRBS data sets
+mated the CpG coverage in the merged groups of two to eight    were discarded.
+ESCs, we simply added the CpG sites only if these CpG sites were
+captured at least once in any one of these single cells. However,
+                                                             Data reproducibility
+when the methylation level of the merged eight single cells was
+computed, only CpG sites that were covered in at least six single-    To estimate the reproducibility of our method, the Pearson corre-
+cell samples with no less than 33 coverage in each single cell were     lation coefficients for all of the samples were computed using the R
+considered. The subsequent statistical computing and graphics   command ‘‘cor’’ with ‘‘pairwise.complete.obs’’ as the value of the
+were performed with customized Perl scripts and R packages    parameter ‘‘use.’’ An unsupervised hierarchical clustering analysis
+(http://www.r-project.org/).                                   was performed using the ‘‘hclust’’ function in R software and was
+                                                                      further integrated with a customized correlation heatmap (Fig. 2B).
+
+Annotation of genomic regions
+                                            5mC and 5hmC immunostaining
+High-density CpG promoter (HCP), intermediate-density CpG
+promoter (ICP), and low-density CpG promoter (LCP) annotations    Zygotes collected from the mouse oviduct ampulla were fixed
+were all taken from the reference by Mikkelsen et al. (2007) with-    with 4% paraformaldehyde (Sigma) for 15 min at room temper-
+out any modifications. In detail, three promoter types were defined     ature and washed three times in PBST, followed by permeabiliza-
+based on the transcription start sites (TSS) of known RefSeq genes.     tion with 0.5% Triton X-100 (Sigma) for 15 min. The DNA was then
+In detail, HCP, which indicated the ‘‘CpG-rich’’ promoters, was    denatured with 4 M HCl for 10 min and neutralized with 100 mM
+identified as having a GC density $0.55 and the observed to     Tris-HCl (pH 8.5) for 15 min at room temperature. The zygotes
+expected CpG ratio (CpG O/E) $ 0.6; promoters with CpG O/E #    were then blocked with 0.1% PBS-BSA (Sigma) overnight at 4°C
+0.4 were classified as LCP; the remaining nonoverlapping pro-    and incubated with anti-5mc antibody (1/200, BIMECY-0500;
+moter populations (0.4 < CpG O/E < 0.6) were classified as ICP. The     Eurogentec) or anti-5-hmC antibody (1/500, 39769; Active Motif)
+annotated repeat elements such as LINEs, SINEs, and LTRs were     for 1 h at 37°C. After washing in PBST three times, the zygotes
+downloaded directly from the RepeatMasker track of the UCSC    were incubated with Alexa Fluor 568 goat anti-mouse IgG (1/500,
+Genome Browser. Other regions such as CGIs, exons, and introns    A-11004; Invitrogen) or donkey anti-rabbit IgG-FITC (1/100,
+were downloaded from the UCSC Genome Browser. Intragenic     sc-2012; Santa Cruz) for 1 h at 37°C. Finally, the zygotes were
+regions were included from the TSS to the transcription termina-    mounted with 5 mg/mL DAPI (Sigma), and fluorescence was de-
+tion sites (TTS), whereas the intergenic regions were defined as the     tected under an inverted fluorescence microscope (Nikon) using
+complement of the intragenic regions.                         an EM-CCD camera. All images were acquired and analyzed using
+                                                                NIS-Elements BR Microscope Imaging Software (Nikon) (Supple-
+                                                             mental Fig. 9B).
+Calculation of CpG site methylation levels
+The methylation level of each single CpG site was estimated as the   Data access
+number of reported Cs (methylated) divided by the total number
+of reported Cs (methylated) or Ts (unmethylated) at the same po-     All sequencing data have been submitted to the NCBI Gene Ex-
+sition of the reference genome. For the single-cell data, we only     pression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) un-
+selected CpG sites that were covered by no less than three reads in     der accession number GSE47343.
+depth for the subsequent analysis, regardless of the amplification
+bias and errors introduced in the preparation of the libraries or   Acknowledgments
+high-throughput sequencing workflow. Theoretically, every cov-
+ered CpG site in our single-cell RRBS method should be defined   We thank C. He, J. Qiao, X.S. Xie, and Y. Huang for their insight-
+digitally as either fully methylated (100%) or unmethylated (0%),     ful discussion and useful assistance. The project was supported
+respectively. Considering the potential amplification and  se-    by grants from the National Basic Research Program of China
+quencing errors, the methylation level of $90% or #10% CpG    (2012CB966704 and 2011CB966303) and the National Natural
+sites was reassigned as fully methylated (100%) or unmethylated     Science Foundation of China (31271543).
+
+
+                                                                 Genome Research  9
+                                                                                              www.genome.org
+
+## Page 11
+
+              Downloaded from genome.cshlp.org on June 10, 2026 . Published by Cold Spring Harbor Laboratory Press
+
+
+
+Guo et al.
+
+
+References                                                    Mayer W, Niveleau A, Walter J, Fundele R, Haaf T. 2000. Demethylation of
+                                                                             the zygotic paternal genome. Nature 403: 501–502.
+                                                                        Meissner A, Gnirke A, Bell GW, Ramsahoye B, Lander ES, Jaenisch R. 2005.
+Bao S, Tang F, Li X, Hayashi K, Gillich A, Lao K, Surani MA. 2009. Epigenetic
+                                                                      Reduced representation bisulfite sequencing for comparative high-
+    reversion of post-implantation epiblast to pluripotent embryonic stem
+                                                                                   resolution DNA methylation analysis. Nucleic Acids Res 33: 5868–5877.
+     cells. Nature 461: 1292–1295.
+                                                                       Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G, Alvarez P,
+Bird A. 2002. DNA methylation patterns and epigenetic memory. Genes Dev
+                                                                  Brockman W, Kim TK, Koche RP, et al. 2007. Genome-wide maps of
+   16: 6–21.
+                                                                       chromatin state in pluripotent and lineage-committed cells. Nature 448:
+Bock C. 2012. Analysing and interpreting DNA methylation data. Nat Rev
+                                                                         553–560.
+   Genet 13: 705–719.
+                                                                  Navin N, Kendall J, Troge J, Andrews P, Rodgers L, McIndoo J, Cook K,
+Booth MJ, Branco MR, Ficz G, Oxley D, Krueger F, Reik W, Balasubramanian
+                                                                           Stepansky A, Levy D, Esposito D, et al. 2011. Tumour evolution inferred
+    S. 2012. Quantitative sequencing of 5-methylcytosine and
+                                                                    by single-cell sequencing. Nature 472: 90–94.
+   5-hydroxymethylcytosine at single-base resolution. Science 336: 934–
+                                                               Okada Y, Yamagata K, Hong K, Wakayama T, Zhang Y. 2010. A role for the
+   937.                                                                               elongator complex in zygotic paternal genome demethylation. Nature
+Chan MM, Smith ZD, Egli D, Regev A, Meissner A. 2012. Mouse ooplasm
+                                                                       463: 554–558.
+    confers context-specific reprogramming capacity. Nat Genet 44: 978–                                                            Okano M, Bell DW, Haber DA, Li E. 1999. DNA methyltransferases Dnmt3a
+   980.                                                                  and Dnmt3b are essential for de novo methylation and mammalian
+Deaton AM, Bird A. 2011. CpG islands and the regulation of transcription.                                                                         development. Cell 99: 247–257.
+   Genes Dev 25: 1010–1022.                                                                 Oswald J, Engemann S, Lane N, Mayer W, Olek A, Fundele R, Dean W, Reik
+Farthing CR, Ficz G, Ng RK, Chan CF, Andrews S, Dean W, Hemberger M,                                                              W, Walter J. 2000. Active demethylation of the paternal genome in the
+    Reik W. 2008. Global mapping of DNA methylation in mouse promoters                                                                mouse zygote. Curr Biol 10: 475–478.
+    reveals epigenetic reprogramming of pluripotency genes. PLoS Genet 4:                                                                           Reik W. 2007. Stability and flexibility of epigenetic gene regulation in
+   e1000116.                                                              mammalian development. Nature 447: 425–432.
+Ferreira J, Carmo-Fonseca M. 1997. Genome replication in early mouse                                                                           Rodriguez-Paredes M, Esteller M. 2011. Cancer epigenetics reaches
+   embryos follows a defined temporal and spatial order. J Cell Sci 110: 889–                                                                      mainstream oncology. Nat Med 17: 330–339.
+   897.                                                                         Santos F, Hendrich B, Reik W, Dean W. 2002. Dynamic reprogramming
+Gomez D, Shankman LS, Nguyen AT, Owens GK. 2013. Detection of histone                                                                                  of DNA methylation in the early mouse embryo. Dev Biol 241: 172–
+    modifications at specific gene loci in single cells in histological sections.                                                                            182.
+   Nat Methods 10: 171–177.                                                                 Smallwood SA, Kelsey G. 2012. Genome-wide analysis of DNA methylation
+Gu H, Bock C, Mikkelsen TS, Ja¨ger N, Smith ZD, Tomazou E, Gnirke A,                                                                                 in low cell numbers by reduced representation bisulfite sequencing.
+   Lander ES, Meissner A. 2010. Genome-scale DNA methylation mapping        Methods Mol Biol 925: 187–197.
+    of clinical samples at single-nucleotide resolution. Nat Methods 7: 133–     Smallwood SA, Tomizawa S, Krueger F, Ruf N, Carli N, Segonds-Pichon A,
+   136.                                                                        Sato S, Hata K, Andrews SR, Kelsey G. 2011. Dynamic CpG island
+Gu H, Smith ZD, Bock C, Boyle P, Gnirke A, Meissner A. 2011a. Preparation        methylation landscape in oocytes and preimplantation embryos. Nat
+    of reduced representation bisulfite sequencing libraries for genome-scale         Genet 43: 811–814.
+  DNA methylation profiling. Nat Protoc 6: 468–481.                     Smith ZD, Gu H, Bock C, Gnirke A, Meissner A. 2009. High-throughput
+Gu TP, Guo F, Yang H, Wu HP, Xu GF, Liu W, Xie ZG, Shi L, He X, Jin SG, et al.          bisulfite sequencing in mammalian genomes. Methods 48: 226–232.
+   2011b. The role of Tet3 DNA dioxygenase in epigenetic reprogramming     Smith ZD, Chan MM, Mikkelsen TS, Gu H, Gnirke A, Regev A, Meissner A.
+   by oocytes. Nature 477: 606–610.                                          2012. A unique regulatory phase of DNA methylation in the early
+Hackett JA, Reddington JP, Nestor CE, Dunican DS, Branco MR, Reichmann       mammalian embryo. Nature 484: 339–344.
+     J, Reik W, Surani MA, Adams IR, Meehan RR. 2012. Promoter DNA        Tang F, Barbacioru C, Wang Y, Nordman E, Lee C, Xu N, Wang X, Bodeau J,
+   methylation couples genome-defence mechanisms to epigenetic             Tuch BB, Siddiqui A, et al. 2009. mRNA-seq whole-transcriptome
+   reprogramming in the mouse germline. Development 139: 3623–3632.          analysis of a single cell. Nat Methods 6: 377–382.
+Hackett JA, Sengupta R, Zylicz JJ, Murakami K, Lee C, Down TA, Surani MA.     Tang F, Barbacioru C, Nordman E, Bao S, Lee C, Wang X, Tuch BB, Heard E,
+   2013. Germline DNA demethylation dynamics and imprint erasure           Lao K, Surani MA. 2011a. Deterministic and stochastic allele specific
+   through 5-hydroxymethylcytosine. Science 339: 448–452.                   gene expression in single mouse blastomeres. PLoS ONE 6: e21208.
+Inoue A, Matoba S, Zhang Y. 2012. Transcriptional activation of              Tang F, Lao K, Surani MA. 2011b. Development and applications of single-
+    transposable elements in mouse zygotes is independent of Tet3-                   cell transcriptome analysis. Nat Methods 8: S6–S11.
+   mediated 5-methylcytosine oxidation. Cell Res 22: 1640–1649.           Toyooka Y, Shimosato D, Murakami K, Takahashi K, Niwa H. 2008.
+Iqbal K, Jin SG, Pfeifer GP, Szabo´ PE. 2011. Reprogramming of the paternal          Identification and characterization of subpopulations in
+   genome upon fertilization involves genome-wide oxidation of                  undifferentiated ES cell culture. Development 135: 909–918.
+    5-methylcytosine. Proc Natl Acad Sci 108: 3642–3647.                    Wossidlo M, Nakamura T, Lepikhov K, Marques CJ, Zakhartchenko V, Boiani
+Jaenisch R, Bird A. 2003. Epigenetic regulation of gene expression: How the       M, Arand J, Nakano T, Reik W, Walter J. 2011. 5-Hydroxymethylcytosine
+   genome integrates intrinsic and environmental signals. Nat Genet 33           in the mammalian zygote is linked with epigenetic reprogramming. Nat
+    (Suppl):245–254.                                           Commun 2: 241.
+Kalisky T, Quake SR. 2011. Single-cell genomics. Nat Methods 8: 311–314.    Wu H, Zhang Y. 2011. Mechanisms and functions of Tet protein-mediated
+Krueger F, Andrews SR. 2011. Bismark: A flexible aligner and methylation          5-methylcytosine oxidation. Genes Dev 25: 2436–2452.
+    caller for bisulfite-seq applications. Bioinformatics 27: 1571–1572.        Xu YN, Cui XS, Tae JC, Jin YX, Kim NH. 2011. DNA synthesis and epigenetic
+Kurimoto K, Yabuta Y, Ohinata Y, Ono Y, Uno KD, Yamada RG, Ueda HR,          modification during mouse oocyte fertilization by human or hamster
+    Saitou M. 2006. An improved single-cell cDNA amplification method for        sperm injection. J Assist Reprod Genet 28: 325–333.
+    efficient high-density oligonucleotide microarray analysis. Nucleic Acids    Yu M, Hon GC, Szulwach KE, Song CX, Zhang L, Kim A, Li X, Dai Q, Shen Y,
+    Res 34: e42.                                                               Park B, et al. 2012. Base-resolution analysis of 5-hydroxymethylcytosine
+Li E. 2002. Chromatin modification and epigenetic reprogramming in             in the mammalian genome. Cell 149: 1368–1380.
+   mammalian development. Nat Rev Genet 3: 662–673.                  Zong C, Lu S, Chapman AR, Xie XS. 2012. Genome-wide detection of single-
+Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, Nery         nucleotide and copy-number variations of a single human cell. Science
+    JR, Lee L, Ye Z, Ngo QM, et al. 2009. Human DNA methylomes at base        338: 1622–1626.
+    resolution show widespread epigenomic differences. Nature 462: 315–
+   322.
+Marusyk A, Almendro V, Polyak K. 2012. Intra-tumour heterogeneity:
+   A looking glass for cancer? Nat Rev Cancer 12: 323–334.                      Received June 6, 2013; accepted in revised form September 16, 2013.
+
+
+
+
+
+10  Genome Research
+       www.genome.org

scrrbs/scRRBS.human_text.txt

@@ -0,0 +1,291 @@
+Single-cell reduced representation bisulfite sequencing
+
+Protocol
+
+scRRBS is a single-cell reduced representation bisulfite sequencing method for DNA methylation profiling. Single-cell genomic DNA is processed in a single-tube reaction to minimize DNA loss before PCR amplification. A single cell is lysed, unmethylated lambda DNA spike-in is added, and genomic DNA is digested with MspI, a CpG-rich restriction enzyme that recognizes CCGG sites. The digested DNA is end-repaired and dA-tailed, premethylated indexed TruSeq adapters are ligated, and the ligated DNA is bisulfite converted. After bisulfite conversion and purification, converted DNA is amplified by two rounds of PCR, size-selected, quality controlled, and sequenced on an Illumina platform.
+
+Key oligo and library-related sequences
+
+The explicit adapter sequences are provided in the protocol table of adapter oligos. All adapter sequences are written 5' to 3'.
+
+1. TruSeq universal adapter
+
+Source sequence:
+
+AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
+
+2. TruSeq adapter, index 1
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACATCACGATCTCGTATGCCGTCTTCTGCTTG
+
+3. TruSeq adapter, index 2
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATGTATCTCGTATGCCGTCTTCTGCTTG
+
+4. TruSeq adapter, index 3
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACTTAGGCATCTCGTATGCCGTCTTCTGCTTG
+
+5. TruSeq adapter, index 4
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACTGACCAATCTCGTATGCCGTCTTCTGCTTG
+
+6. TruSeq adapter, index 5
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACACAGTGATCTCGTATGCCGTCTTCTGCTTG
+
+7. TruSeq adapter, index 6
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG
+
+8. TruSeq adapter, index 7
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACCAGATCATCTCGTATGCCGTCTTCTGCTTG
+
+9. TruSeq adapter, index 8
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACACTTGAATCTCGTATGCCGTCTTCTGCTTG
+
+10. TruSeq adapter, index 9
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACGATCAGATCTCGTATGCCGTCTTCTGCTTG
+
+11. TruSeq adapter, index 10
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACTAGCTTATCTCGTATGCCGTCTTCTGCTTG
+
+12. TruSeq adapter, index 11
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACGGCTACATCTCGTATGCCGTCTTCTGCTTG
+
+13. TruSeq adapter, index 12
+
+Source sequence:
+
+GATCGGAAGAGCACACGTCTGAACTCCAGTCACCTTGTAATCTCGTATGCCGTCTTCTGCTTG
+
+14. QP1 primer
+
+Source sequence:
+
+AATGATACGGCGACCACCGA
+
+15. QP2 primer
+
+Source sequence:
+
+CAAGCAGAAGACGGCATACGA
+
+
+
+Step-by-step library generation
+
+Step 1. Single-cell lysis
+
+Input substrate:
+
+* one isolated cell
+
+Added oligos/reagents:
+
+* lysis buffer
+* protease
+
+Molecular event:
+The single cell is lysed in a small-volume reaction, releasing genomic DNA in the same tube used for downstream library construction.
+
+Product structure:
+
+* released genomic DNA from one cell
+
+Step 2. Lambda spike-in and MspI digestion
+
+Input substrate:
+
+* released genomic DNA
+
+Added oligos/reagents:
+
+* unmethylated lambda DNA spike-in
+* MspI restriction enzyme
+* Tango buffer
+
+Molecular event:
+Unmethylated lambda DNA is added as a bisulfite-conversion control. MspI digests genomic DNA at CCGG sites, generating reduced-representation genomic fragments enriched for CpG-rich regions.
+
+Product structure:
+
+MspI-digested genomic DNA fragments
+
+Step 3. End repair and dA-tailing
+
+Input substrate:
+
+* MspI-digested genomic DNA fragments
+
+Added oligos/reagents:
+
+* Klenow fragment, exo-
+* dATP
+* dCTP
+* dGTP
+* Tango buffer
+
+Molecular event:
+Digested DNA fragments are end-repaired and dA-tailed to prepare them for ligation to TruSeq adapters.
+
+Product structure:
+
+* end-repaired, dA-tailed MspI genomic fragments
+
+Step 4. Adapter ligation
+
+Input substrate:
+
+* end-repaired, dA-tailed MspI genomic fragments
+
+Added oligos/reagents:
+
+* premethylated indexed TruSeq adapters
+* T4 DNA ligase
+* ATP
+* Tango buffer
+
+Molecular event:
+Premethylated indexed TruSeq adapters are ligated to the dA-tailed DNA fragments. The indexed adapter contains a 6-base sample index. The adapter cytosines are methylated so that they are protected during bisulfite conversion.
+
+Product structure:
+
+TruSeq universal adapter
+
+* MspI genomic DNA fragment
+* indexed TruSeq adapter
+
+Step 5. Bisulfite conversion
+
+Input substrate:
+
+* adapter-ligated genomic DNA fragments
+
+Added oligos/reagents:
+
+* bisulfite conversion reagent
+* tRNA carrier during purification
+
+Molecular event:
+Adapter-ligated DNA is bisulfite converted. Unmethylated cytosines in genomic DNA are converted, while methylated cytosines remain protected. The converted DNA is then purified after bisulfite treatment.
+
+Product structure:
+
+* bisulfite-converted adapter-ligated DNA fragments
+
+Step 6. First-round PCR enrichment
+
+Input substrate:
+
+* purified bisulfite-converted adapter-ligated DNA
+
+Added oligos/reagents:
+
+* QP1 primer
+* QP2 primer
+* PfuTurbo Cx hotstart DNA polymerase
+* dNTP mix
+
+Molecular event:
+The bisulfite-converted library molecules are amplified using QP1 and QP2 primers. PfuTurbo Cx is used because it is resistant to uracil stalling.
+
+Product structure:
+
+* first-round PCR-amplified scRRBS library molecules
+
+Step 7. Second-round PCR enrichment
+
+Input substrate:
+
+* first-round PCR product
+
+Added oligos/reagents:
+
+* QP1 primer
+* QP2 primer
+* Phusion high-fidelity PCR master mix or KAPA HiFi HotStart ReadyMix
+
+Molecular event:
+A second PCR round further enriches the scRRBS library.
+
+Product structure:
+
+* amplified scRRBS library molecules
+
+Step 8. Size selection and sequencing
+
+Input substrate:
+
+* amplified scRRBS library molecules
+
+Added oligos/reagents:
+
+* native polyacrylamide TBE gel
+* AMPure XP beads
+
+Molecular event:
+Amplified DNA fragments are size-selected, typically keeping fragments in the 200–700 bp range. The final library is quantified, quality controlled, and sequenced as paired-end Illumina libraries.
+
+Product structure:
+
+* final size-selected scRRBS sequencing library
+
+Final canonical library structure
+
+Simplified segment-level structure:
+
+TruSeq universal adapter + bisulfite-converted MspI genomic insert + indexed TruSeq adapter
+
+Source-visible sequence-level structure with generic genomic insert:
+
+AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
++ bisulfite-converted MspI genomic insert
++ GATCGGAAGAGCACACGTCTGAACTCCAGTCAC
++ 6-bp sample index
++ ATCTCGTATGCCGTCTTCTGCTTG
+
+
+Sequencing read interpretation:
+
+* Read 1 and Read 2 sequence the bisulfite-converted reduced-representation genomic insert from opposite adapter ends.
+* The indexed TruSeq adapter provides a 6-bp sample index.
+* The library is used for DNA methylation profiling after bisulfite-aware alignment and methylation calling.
+
+Human-curation notes
+
+1. The protocol table explicitly lists the TruSeq universal adapter and TruSeq indexed adapters 1–12. All adapter sequences are written 5' to 3'.
+2. The indexed TruSeq adapters contain 6-base sample indexes.
+3. The protocol notes that the last two bases of the TruSeq universal adapter should be phosphorothioated to avoid nuclease cleavage of the T overhang.
+4. The protocol notes that TruSeq indexed adapters 1–12 should be 5'-terminal phosphorylated for ligation to inserted DNA fragments.
+5. QP1 and QP2 primer sequences are explicitly printed in the protocol materials section and should be included in the sequence list.
+6. The one-tube reaction integrates cell lysis, MspI digestion, end repair/dA-tailing, adapter ligation, and bisulfite conversion before DNA purification.
+7. The final genomic insert should be interpreted as a bisulfite-converted MspI genomic insert, not ordinary unconverted gDNA.
+8. The exact genomic insert sequence is sample dependent and should not be fixed.