Figure 3.

The history of chromosomal instability over tumor evolution. A, The number of copy number events required to reach the final copy number state from the most recent common ancestor from a collection of tumor cells with different ploidy states from a single undifferentiated sarcoma. The number of events is normalized for ploidy and statistical significance is calculated with a Mann–Whitney U test. Ψ denotes population ploidy as determined by FACS. B, The normalized rate of gains relative to the WGD timing across cancer types. Statistical significance is calculated with a permutation test and 95% confidence intervals are calculated by bootstrapping over samples. C, Histograms of gain timing in a random selection of tumors with a WGD. Samples are ordered by independent gain timing from left to right and WGD timing from top to bottom. D, Proportion of samples with gains post-WGD for WGD tumors and a cohort of control non-WGD tumors with a pseudo-WGD timing randomly sampled from WGD tumors with the same cancer type. Statistical significance is calculated with a permutation test and 95% confidence intervals are calculated by bootstrapping over samples. E, Distribution of the mean mutation time between the timing of all independent gains and the median WGD timing for each genome duplicated sample. Two cohorts are displayed, one with the correct WGD timing and another where the WGD timing is permuted between samples of the same cancer type. Statistical significance calculated by Mann–Whitney U test. F, Proportion of genome gained before and after WGD against WGD timing for genome-doubled tumors. G, Proportion of genome lost before and after WGD against WGD timing for genome-doubled tumors. H, Proportion of tumors with clonal gains identified as occurring in a punctuated burst, or uninformative where the number of gains was too low to classify, split by WGD status. I, Proportion of punctuated gains occurring in WGD samples, classified by whether they occurred pre- or post-WGD. **, P < 0.01; ***, P < 0.001.