Estimation of the admixture of APOBEC3A/B mutations from the leading vs. lagging strand asymmetry

Previously, using non-tissue-specific replication timing data as in the current analysis of heritable mutations, we found that about two-thirds of APOBEC3A/B mutations in cancer occurred on the lagging strand (Seplyarskiy et al. 2016). By using tissue-matched data on replication timing for analyses of asymmetry in cancers (MCF-7 for breast cancers, and IMR90 for lung cancers), we increase the observed level of asymmetry (Supplemental Fig. S10) compared with our previous study and studies by other groups (Haradhvala et al. 2016; Morganella et al. 2016; Seplyarskiy et al. 2016). Therefore, in our non-tissue-matched data, the observed ratio of mutational frequencies on the lagging and on the leading strands (strand bias) s is

where x is the fraction of APOBEC3A/B-induced mutations among all mutations; 0.67x and 0.33x are the fractions of APOBEC3A/B-induced mutations on the lagging and on the leading strand, respectively, with the coefficients 0.67 and 0.33 estimated from cancer data; and the coefficient 0.5 corresponds to the equal distribution of non-APOBEC3A/B-induced mutations between leading and lagging strands. Therefore,

so that s = 1.15 estimated for heritable mutations implies an admixture of APOBEC3A/B-induced mutations of 21%.