Supplementary Figure 4: Characteristics of HLA mutations detected by POLYSOLVER across 7,930 samples.

(a) Allelic frequencies of all 298 detected HLA somatic changes. The median allele fraction across somatic changes was 33% (interquartile range: 16–58%). Most of these mutations are likely heterozygous. (b) Frequency of HLA mutations in samples. 240 of 266 (90.2%) samples with HLA mutations only had a single somatic event, 20 had two and 6 samples (4 colon, 1 stomach and 1 uterine) had 3 distinct HLA mutations. (c) Frequency of cases per recurrently mutated site. 57 of 64 recurrently mutated sites were defined as recurrent on the basis of 2 to 4 specimens across samples with a mutation at the same site. Residues 25, 299, 7 and 209 were found to be highly recurrent with 7, 9, 11 and 24 distinct individuals harboring mutations at these two positions respectively. (d) Length-normalized distribution of HLA mutations across functional domains. A strong preference of potentially loss-of-function events (nonsense, frameshift indels, splice site mutations) for exon 1 is observed.