Fig. 4

Negative selection pressure on SV-derived neoantigens. a Given a shared neo-peptide from a gene, samples that bear the neo-peptide are retrieved and grouped to neoantigen and neo-peptide samples according to whether they could present the neo-peptide or not. The ratio between the average expressions of the gene in the neoantigen and neo-peptide samples is shown. b Proportions of neoantigenic SVs in oncogenic SVs and passenger SVs (two-sided Fisher’s exact test). c Proportions of TMPRSS2-ERG and KIAA1549-BRAF fusions that result in neoantigens and do not result in neoantigens. d Comparison of patient-level ITH of SV-, SNV-, and indel-derived neoantigens (two-sided Wilcoxon rank-sum test). Boxplot hinges represent 25th to 75th percentiles, and central lines represent median values; violin plots refer to the kernel probability densities. e 2D density plots showing the ITH of SV-derived neoantigens versus that of SNV-derived neoantigens in lung squamous carcinoma (Lung-SCC, left) and prostate adenocarcinoma (Prost-AdenoCA, right). f Distributions of odds ratios of generating neoantigens from subclonal and clonal SVs in different patients. The odds ratios are log-transformed. Red bars indicate median values. Cancer types with median odds ratio significantly deviated from 1 are labeled (one-sample Wilcoxon rank-sum test)