Fig. 4

Sequence depth analysis and library complexity evaluation. It is important to know that it is not meaningful to perform saturation analysis of sequencing depth or library complexity for over-transposed ATAC-seq assays. (a) Total peak number-based saturation analysis of sequencing depth for SRR891270. Sequenced fragments in the filtered BAM file (called effective fragments here) are subsampled to get 10%, 20%, 30%…, 80% and 90% of total effective fragments. Broad peaks were called for each subsample and the full dataset using MACS2. The numbers of significant peaks (FDR ≤ 0.05) are plotted against the corresponding numbers of effective fragments. A smooth curve is fitted by using the geom_loess function in the ggplot2 package. The gray band shows the 95% confidence interval of the predicted peak numbers. (b) Total peak width-based saturation analysis of sequencing depth for SRR891270. The same procedure is used to fit the saturation curve except that the total width of significant peaks (FDR ≤ 0.05) for each subsample and the full dataset is used. (c) Library complexity analysis results for SRR891269-SRR891271, three biological replicates using 50 K cells, and for SRR891272 and SRR891274, two biological replicates using 500 cells. Number of distinct fragments was estimated for each given number of putative sequenced fragments free of mitochondrial reads