Supplementary Figure 4: YTCA enrichment versus RTCA enrichment for samples that were significantly enriched for TCA.
All samples are binned by quartile of TCA enrichment. Unfilled, gray-bordered symbols denote samples where the ratio of YTCA to RTCA mutation is not statistically different from random, regardless of TCA enrichment. (a–e) Whole-genome analysis results for (a) colorectal, (b) multiple myeloma, (c) melanoma, (d) thyroid and (e) uterine endometrial cancer cohorts are shown. Note that the apparently high presence of TCA APOBEC signature–enriched samples in melanoma is likely artifactual, due to considerable overlap with the mutation signature for ultraviolet light (C→T mutations at dipyrimidines) (discussed in ref. 27). No samples are significantly enriched for TCA among the other five cohorts (glioblastoma multiforme, kidney chromophobe, kidney renal clear cell, lower-grade glioma and prostate) with whole-genome mutation data. (f–k) Results for whole-exome mutation data from (f) bladder, (g) breast, (h) cervical, (i) head and neck, (j) lung adenocarcinoma and (k) lung squamous cell cohorts are shown. Most samples fail to pass statistical filtering for a non-random ratio of YTCA mutations to RTCA mutations, because of the low mutation counts per sample and the large numbers of samples within each cohort. The few samples that do pass statistical filtering all had high mutation counts and A3A-like signatures. These results show that, in low–APOBEC mutagenesis genomes or high–APOBEC mutagenesis exomes, samples that do pass both levels of statistical filtering tend to have high TCA enrichments and A3A-like signatures (YTCA > RTCA enrichment).
