Fig. 2

Performance of CEL-Seq2 compared to the original CEL-Seq. In total, 24, 20, and 72 single fibroblasts were analyzed by CEL-Seq, CEL-Seq2, and CEL-Seq2(C1), respectively. a For each of the 92 ERCC spike-ins the average observed and expected molecules across the examined cells is plotted. The lines were fitted using a linear model and the efficiency was computed as the y-intercept. b Comparing the mean number of transcripts identified per cell between the methods. c Comparing the mean number of genes identified per cell between the methods. d Coefficient of variation for the original CEL-Seq and CEL-Seq2. Each dot indicates a gene’s squared coefficient of variation. e A comparison between the squared coefficients of variation for the detected ERCC spike-ins. Error bars indicate standard error