Fig. 2

Neoepitope predictions in TCGA across disease sites and HLA alleles. a. Number of total and putatively novel-binding predicted neoepitopes in each disease group from TCGA. The total number of neoepitopes (gray) for each patient in each disease group, shown in order of decreasing median neoepitope burden, was determined using pVAC-Seq. Novel binding (red) are the subset of neoepitopes which displayed a putatively novel binding change (see Methods). Vertical lines separate TCGA disease sites. Outliers have been removed for clarity. On average, 20.3% of a patient’s neoepitopes were novel-binding. A TCGA disease abbreviation key is available in Additional file 1: Table S1. b. Putatively novel binding predicted neoepitopes across HLA alleles in TCGA. Top: number of neoepitopes with a novel binding change in TCGA for each HLA allele studied, colored green, blue, and yellow according to HLA allele types (A, B, and C, respectively). Bottom: average population frequency for each HLA allele studied, colored as per Top pane. Alleles with > 10% frequency in the population are labeled. There was no relationship between allele frequency and number of putatively novel binding neoepitopes (Pearson’s product-moment correlation of 0.1, p = 0.1)