Extended Data Figure 2: Validation of the subclonal hierarchies in A22.

The primary means of validation was a deep sequencing validation experiment that included selected substitutions and indels from each sample, as described in Extended Data Table 2 and Supplementary Information section 2b. In addition, indels and rearrangements identified in WGS represent data sets orthogonal to the substitution data from which the subclones were identified. The subsets of samples in which validated substitutions, indels and rearrangements are found correlate strongly with the subclonal clusters identified from the clustering of substitutions from WGS, providing support for the existence of these subclones. a, b, For each patient, hierarchical clustering of the variant allele fraction (VAF) was performed separately for substitutions (a) and indels (b). VAFs are represented as a heat map with deeper shades of red indicating a higher proportion of reads reporting the mutant allele. Above each heat map, mutations are colour-coded according to the subclone they were assigned to by Dirichlet process clustering of WGS data in the case of substitutions or by VAF for indels. Indels that could not be assigned to any cluster are annotated with black. For A22, additional samples not subject to WGS were included in the validation experiment. c, For these patients the phylogenetic tree from Fig. 2 was modified to incorporate these additional samples. d–f, Number of substitutions assigned to each subclone (d) and numbers of indels (e) and rearrangements (f) present in different subsets of samples are plotted as bar charts. g, VAFs from whole-genome sequencing and validation data, plotted as scatter plots, are very highly correlated. h, Subclone colour key.