Fig. 4: An optimized T7 promoter boosts single-cell RNA-Sequencing.

a Addition of an AT-rich upstream promoter-flanking region enhances the activity of the T7 promoter at low template concentrations. A 410 nucleotide long RNA was in vitro transcribed for 2 h (1 nanogram template), or 15 h (1 picogram template) in the absence or presence of an AT-rich upstream element as indicated. Shown is the relative RNA yield. The start of the promoter sequence was either located at position 73 (left bars), or at position 6 of the DNA template (middle and right bars). Error bars represent the standard deviation of triplicate experiments. b CEL-Seq2 was performed from single K562 cells using the indicated DNA sequences. After reverse transcription and second strand synthesis, cDNA from 10 cells was pooled and in vitro transcribed for 15 h. Purified aRNA was fragmented and quantified on a Tapestation (Agilent). Error bars represent the standard deviation from triplicate experiments. c Linear amplification of cDNA in single cells with an optimized T7 promoter (CEL-Seq+) significantly increased the number of detected genes in single-cell RNA-Sequencing (9749 genes per cell on average with new T7 promoter (n = 24 cells), 8281 genes per cell on average with conventional T7 promoter (n = 14 cells). d Linear amplification with an optimized T7 promoter (CEL-Seq+) significantly increased the number of detected molecules in single-cell RNA-Sequencing (85,066 unique molecular identifiers (UMIs)/transcripts per cell on average with new T7 promoter, 53541 UMIs per cell on average with conventional T7 promoter). Statistical analysis was performed using the Mann–Whitney Wilcoxon test. e Average UMI count per gene in CEL-Seq2 and CEL-Seq+. Shown are all genes with more than 1 average UMI per cell in both assays (n = 7904). Genes are independently sorted by expression rank. Fourteen cells from CEL-Seq+ were randomly chosen for comparison with 14 cells from CEL-Seq2. Error bars show the standard deviation between cells. f Coefficient of variation for UMI counts from CEL-Seq2 and CEL-Seq+ for the genes shown in (e). g Genes from deep sequenced bulk K562 RNA-Seq²⁰ were sorted into quartiles by expression level. Shown are the numbers of genes from the indicated bulk quartiles that were detected in individual cells by CEL-Seq2 or CEL-Seq+. h Shown are the numbers of genes detected in individual cells that are differentially expressed throughout CML disease progression²⁰, or with GO association “DNA binding transcription factor”. Whiskers reach to 1.5× IQR away from the 1st/3rd quartile.