Pervasive tissue specificity of driver genes revealed by mutational analysis of 265 cancer types

Defining genes that are somatically mutated in different cancer types is a central goal of cancer genetics. Nevertheless, traditional definitions of "driver" genes are biased toward common cancer types and tend to overlook genes that might be specific to rarer subtypes. We developed a statistical framework that defines genes enriched for functional somatic mutations in primary cancers to quantify incidence and tissue specificity for each gene. By applying this framework to the AACR GENIE v18.0 dataset, we identified 165 genes significantly mutated in at least one subtype. We mined this dataset to derive tissue specificity scores for all 165 genes, demonstrating that tissue specificity is the norm, not the exception. We also found that oncogenes with restricted expression across normal tissues tend to exhibit higher tissue-specific mutation patterns in cancer. We anticipate that the resources developed in this study will be useful for cancer research and clinical oncology.