Fig. 6.

Comparison of sequencing coverage on clone (for TNBC16) and subclone (for CRC28) level. Compared were the sites shared in the output of DelSIEVE, Sequenza [43], and Ginkgo [44], if available. For Sequenza and Ginkgo, sites were divided into two groups with copy number (CN) $<2$ and $\ge 2$. For DelSIEVE, sites were also divided into two groups, one with deletions, the other copy neutral. Sequencing coverage transformed with $logp1$ across all cells in the clone or subclone at all sites were plotted for reference. In each group, the violin and the box plots matched the color of the method. The total number of data points in each group was marked with n on the horizontal axis. Box plots comprise medians, boxes covering the interquartile range (IQR), and whiskers extending to 1.5 times the IQR below and above the box. Within- and between-group comparisons were conducted between CN $<2$ and $\ge 2$ of Sequenza and Ginkgo, between deletions and copy neutral of DelSIEVE, and between deletions of DelSIEVE and CN $<2$ of Sequenza and Ginkgo. Each comparison was conducted on the sequencing coverage on the original scale, showing the result of two-sided Mann-Whitney U test, with the p value corrected by Holm–Bonferroni method, and the absolute value of the effect size (Cohen’s d). Comparison of sequencing coverage for all cells in TNBC16 (a) as well as in TP (b), TC (c), and TD (d) subclones in CRC28