Extended Data Fig. 3: Positive selection of HLA-I genes.

a) Needle plots representing the pan-cancer distribution of somatic mutations along the HLA-A, HLA-B and HLA-C protein sequences in the primary (PCAWG) dataset. Mutations are colored according to the consequence type. Rectangles represent the Pfam domains. b) Correlation of LOH of HLA-I rates (y-axis) with the mean genome-wide LOH rates (x-axis) across the metastatic (left) and primary (right) datasets. Each dot represents a cancer type. Horizontal lines represent genome-wide LOH standard deviation across samples from each cancer type. The regression and 95% confidence intervals of the linear regression are represented as a solid line and the adjacent shaded area, respectively. Confidence intervals are calculated using 1,000 bootstraps. P-values are obtained from a two-sided test of no-correlation. c) Distribution of non-focal LOH events along the autosomes in the primary kidney chromophobe cancer cohort. X-ticks represent the chromosomal starting position. Dashed horizontal lines represent the expected mean after randomization. Vertical dashed lines highlight the HLA-I genomic location. CDS pos, coding sequence position. Std., standard deviation.