20/20+: Ratiometric prediction of cancer driver genes

Author:Collin Tokheim
Contact:ctokhei1 AT alumni.jh.edu
Source code:GitHub
Q&A:Biostars (tag: 2020+)

Next-generation DNA sequencing of the exome has detected hundreds of thousands of small somatic variants (SSV) in cancer. However, distinguishing genes containing driving mutations rather than simply passenger SSVs from a cohort sequenced cancer samples requires sophisticated computational approaches. 20/20+ integrates many features indicative of positive selection to predict oncogenes and tumor suppressor genes from small somatic variants. The features capture mutational clustering, conservation, mutation in silico pathogenicity scores, mutation consequence types, protein interaction network connectivity, and other covariates (e.g. replication timing). Contrary to methods based on mutation rate, 20/20+ uses ratiometric features of mutations by normalizing for the total number of mutations in a gene. This decouples the genes from gene-level differences in background mutation rate.

Contents:

Releases

Citation

Collin J. Tokheim, Nickolas Papadopoulos, Kenneth W. Kinzler, Bert Vogelstein, and Rachel Karchin. Evaluating the evaluation of cancer driver genes. PNAS 2016 ; published ahead of print November 22, 2016, doi:10.1073/pnas.1616440113