Extended Data Fig. 5: Comparison of PIK3CA MMs among cohorts.

a, Proportion of PIK3CA mutations according to type, position and amino-acid change in recurrently mutated cancer types (defined as those with 20 or more hotspot/functional mutations) in primary samples from the TCGA, FM and GENIE cohorts. b, Distribution of mutations and fraction of MM⁺ mutations for each hotspot/functional position. Asterisks indicate major positions (in which 10% or more of mutations were present in any of the recurrently mutated cancer types). The horizontal blue dotted lines represent the mean values of major positions. c, Number of MMs according to mutational combinations (n = 407). Red and black circles indicate, respectively, observed and expected (median with 95% confidence intervals) values. Representative combinations and observed numbers are also shown. d, Significant pairwise associations (q < 0.01) with observed/expected ratios among functional domains (n = 471). Orange and blue colours depict co-occurring (observed number of MMs significantly higher than expected) and mutually exclusive (lower than expected) associations. c, d, Two-sided simulation test (n = 10,000) with Benjamini–Hochberg correction. e, Major–major combinations on different alleles within close proximity (E542–E545). f, Proportion of MMs showing concordant or discordant allele frequencies according to mutational combinations. g, Order of major and minor hotspot mutations in major–minor combinations showing discordant allele frequencies (n = 54). Two-sided binomial test. h, Proportion of mutations with APOBEC signature (C-to-G/T at TpCpX trinucleotides) according to hotspot/functional position in PIK3CA in the discovery cohort. i, Proportion of mutations with APOBEC signature according to MM status in all coding (left) and PIK3CA (right) mutations in the discovery cohort. b, f, h, i, Two-sided Fisher’s exact test. Examined numbers are shown in parentheses.