Abstract
BACKGROUND: Cancers are caused by the acquisition of somatic mutations. Numerous efforts have been made to characterize the key driver genes and pathways in glioma, however, the etiology of glioma is still not completely known. This study was implemented to characterize driver genes in glioma independently of somatic mutation frequencies. METHODS: Driver genes and pathways were predicted by OncodriveCLUST, OncodriveFM, Icages, Drgap and Dendrix in glioma using 31,958 somatic mutations from TCGA, followed by an integrative characterization of driver genes. RESULTS: Overall, 685 driver genes and 215 driver pathways were determined by the five tools. FSTL5, HCN1, TMEM132D, TRHDE and KRT222 showed the strongest expression correlation with other genes in the co-expression network of glioma tissues. ST6GAL2, PIK3CA, PIK3R1, TP53 and EGFR are at the core of the protein-protein interaction network. 133 driver genes were up-regulated and associated to poor prognosis, 43 driver genes were down-regulated and related to favorable clinical outcome in glioma patients. The driver genes such as MSH6 and RUNX1T1 might serve as candidate prognostic biomarkers and therapeutic targets in glioma. CONCLUSIONS: The set of new cancer genes and pathways sheds insights into the tumorigenesis of glioma and paves the way for developing driver gene-targeted therapy and prognostic biomarkers in glioma.