Fig. 2. Timing of point mutations shows that recurrent driver gene mutations occur early.
a, Top, distribution of point mutations over different mutation periods in n = 2,778 samples. Middle, timing distribution of driver mutations in the 50 most recurrent lesions across n = 2,583 white listed samples from unique donors. Bottom, distribution of driver mutations across cancer types; colour as defined in the inset. b, Relative timing of the 50 most recurrent driver lesions, calculated as the odds ratio of early versus late clonal driver mutations versus background, or clonal versus subclonal. Error bars denote 95% confidence intervals derived from bootstrap resampling. Odds ratios overlapping 1 in less than 5% of bootstrap samples are considered significant (coloured). The underlying number of samples with a given mutation is shown in a. c, Relative timing of TP53 mutations across cancer types, as in b. The number of samples is defined in the x-axis labels. d, Estimated number of unique lesions (genes) contributing 50% of all driver mutations in different timing epochs across n = 2,583 unique samples, containing n = 5,756 driver mutations with available timing information. Error bars denote the range between 0 and 1 pseudocounts; bars denote the average of the two values. NA, not applicable; NS, not significant.