Mutation loads and cancer signatures. Numbers of SNVs per mbp in individual samples in WES (A) and targeted sequencing (B). Samples were categorized into AP/BP (WES, n = 7; panel, n = 12) and CP (WES, n = 12; panel, n = 31) cases. CP cases with optimal (WES, n = 4; panel, n = 16), suboptimal (WES, n = 3; panel, n = 10), and poor (WES, n = 4; panel, n = 3) responses are marked with green, orange, and red, respectively. CP samples with missing response data (WES, n = 1; panel, n = 2) are marked with gray. Scatterplots compare mutation load per mbp and age in WES (C) and targeted sequencing (D) data sets and mutation load per mbp and Sokal score in WES (E) and targeted sequencing (F). Correlations between variables were assessed using Pearson's correlation coefficient. (G) Normalized weights of trinucleotide signatures identified in AML (n = 11), AP/BP (n = 19), and CP (n = 39) cases. CP samples were also subset to optimal (n = 19), suboptimal (n = 10), and poor (n = 7) responders. Variants were pooled across WES and panel sequencing cases. Weights of most frequent signatures in each cancer type are shown across cancers as separate signatures. Summed weight of other signatures is represented by the category “other.” (H) Correlation matrix showing Spearman correlation between normalized weights of signatures of AML, AP/BP, CP, and CP subsets calculated using WES, panel, or combined WES and panel data. Blue dot color indicates positive correlations and red negative correlations as expressed in r values. The dot size represents −log10 P values. Nonsignificant correlations (P > .05) were removed. *P < .05; ****P < .0001. Diagnostic samples from patients #22, #33, #41, #42 and follow-up sample from patient #42 were excluded from the analysis.