Extended Data Fig. 3: The influence of intertumoral heterogeneity on the discordance of genome-wide somatic profiles between paired biopsies.

a, Boxplot of the fraction of genomic alterations private to the first biopsy (lost alterations, as a fraction of the total number of alterations in the first biopsy), stratified for pairs obtained from the same lesion (n = 83 pairs) or pairs obtained from different lesions (n = 92 pairs). Different classes of genomic alterations are plotted separately: nucleotide substitutions (all, clonal, subclonal), indels (all, clonal, subclonal), structural variants (SVs). Clonal alterations have a probability of clonality above >95%, whereas subclonal alterations have a probability of subclonality >95%. Box center lines, box ranges, whiskers, and dots indicate medians, quartiles, 1.5 interquartile ranges, and outliers, respectively. Two-sided Wilcoxon rank sum test-based P-values are shown (red for significant associations). Only paired biopsies for which it was known whether or not the two biopsies were obtained from the same lesion were used for this analysis (n = 175). b, As A, but the y-axis represents the fraction of alterations private to the second biopsy (gained alterations, as a fraction of the total number of alterations in the second biopsy). c, As A, but the y-axis indicates the Pearson correlation coefficient for the copy number profiles of biopsy pairs.