Fig. 2. Signature profiling in stage IV colorectal cancer (CRC).

a The number of mutations fitted to each signature is shown for plasma WGS data from healthy controls (n = 19, shown in red) and individuals with stage IV CRC (n = 16, shown in blue). Background subtraction was performed, and SNPs were retained (“Methods”). The potential etiology of each signature is listed^12,45. Boxplots represent the median, upper and lower quartiles and whiskers indicate 1.5× IQR. One-sided Wilcoxon tests were performed, and Benjamini–Hochberg (BH)-corrected p values (q) are shown. APOBEC apolipoprotein B mRNA-editing enzyme. b Boxplots of aging and MSI signature contributions in healthy individuals (n = 19) and patients with CRC (n = 16). Mutation counts are background-subtracted (“Methods”). One-sided Wilcoxon tests were performed, and adjusted p values (q) are shown. Boxplots represent the median, upper and lower quartiles, and whiskers indicate 1.5× IQR. c One-sided Pearson correlations between tumor fraction and background-subtracted mutation count for aging and MSI signatures are shown for healthy individuals (n = 19, shown in red) and patients with CRC (n = 16, shown in blue). Adjusted p values (q) are shown. The gray shaded area indicates the 95% CI of the fitted linear model. d One-sided Pearson correlations between tumor mutation burden (TMB) and mutation count per signature are shown for each patient (n = 16). Adjusted p values (q) are shown. The gray shaded area indicates the 95% CI of the fitted linear model. e Heatmap of selected aging and MSI signatures detected in CRC plasma samples with 95% specificity (n = 16, “Methods”). Detected signatures in each sample are indicated in red, non-detected signatures are shown in blue. Samples are annotated with ichorCNA ctDNA fraction and microsatellite instability status (MSI)²¹. Source data are provided as a Source data file. MSS microsatellite stable.