Fig. 2

Most mutational signatures exhibit a significant replication strand asymmetry and/or correlation with replication timing. a The difference of matching and inverse exposure is computed for each sample and signature. For each signature, the median value of these differences (in samples exposed to this signature) is plotted against -log₁₀ q-value (signtest of strand asymmetry per sample; with Benjamini-Hochberg correction). b Percentage of samples that have higher matching than inverse exposure to the signature denoted above/below each bar. c Correlation of exposures with replication timing. The 20-kbp replication domains were divided into four quartiles by their average replication timing (early-replicated in the first quartile, late-replicated in the last quartile) and exposures to signatures were computed in each quartile. Median slope of correlation with the replication timing is plotted on the x-axis, i.e., values on the right denote more mutations in late-replicated regions, values on the left reflect more mutations in early replicating regions. The y-axis represents significance of the correlation of signature with replication timing in individual samples (signtest of correlation slope per sample; with Benjamini-Hochberg correction). d Percentage of samples with a positive correlation of replication timing with exposure to the signature denoted above/below each bar