PT - JOURNAL ARTICLE AU - Jordan Anaya AU - Brian J. Reon AU - Wei-Min Chen AU - Stefan Bekiranov AU - Anindya Dutta TI - A pan-cancer analysis of prognostic genes AID - 10.1101/030924 DP - 2015 Jan 01 TA - bioRxiv PG - 030924 4099 - http://biorxiv.org/content/early/2015/12/23/030924.short 4100 - http://biorxiv.org/content/early/2015/12/23/030924.full AB - Numerous studies have identified prognostic genes in individual cancers, but a thorough pan-cancer analysis has not been performed. In addition, previous studies have mostly used microarray data instead of RNA-SEQ, and have not published comprehensive lists of associations with survival. Using recently available RNA-SEQ and clinical data from the The Cancer Genome Atlas for 6,495 patients, we have investigated every annotated and expressed gene’s association with survival across 16 cancer types. The most statistically significant harmful and protective genes were not shared across cancers, but were enriched in distinct gene sets which were shared across certain groups of cancers. These groups of cancers were independently reconstructed by unsupervised clustering of Cox coefficients (a measure of association with survival) for individual genes or for gene programs. This analysis has revealed unappreciated commonalities among cancers which may provide insights into cancer pathogenesis and rationales for co-opting treatments between cancers.