Genome-wide analysis of non-coding regulatory mutations in cancer

Cancer primarily develops due to somatic alterations in the genome. Advances in sequencing have enabled large-scale sequencing studies across many tumor types, emphasizing discovery of alterations in protein-coding genes. However, the protein-coding exome comprises less than 2% of the human genome. Here, we analyze complete genome sequences of 863 human tumors from The Cancer Genome Atlas and other sources to systematically identify non-coding regions that are recurrently mutated in cancer. We utilize novel frequency and sequence-based approaches to comprehensively scan the genome for non-coding mutations with potential regulatory impact. We identified recurrent mutations in regulatory elements upstream of PLEKHS1, WDR74, and SDHD, as well as previously identified mutations in the TERT promoter. SDHD promoter mutations are frequent in melanoma and associated with reduced gene expression and poor patient prognosis. The non-protein-coding cancer genome remains widely unexplored and our findings represent a step towards targeting the entire genome for clinical purposes.