作者: Raunak Shrestha , Ermin Hodzic , Jake Yeung , Kendric Wang , Thomas Sauerwald
DOI: 10.1007/978-3-319-05269-4_23
关键词:
摘要: A key challenge in cancer genomics is the identification and prioritization of genomic aberrations that potentially act as drivers cancer. In this paper we introduce HIT'nDRIVE, a combinatorial method to identify aberrant genes can collectively influence possibly distant "outlier" based on what call "random-walk facility location" RWFL problem an interaction network. differs from standard location by its use "multi-hitting time", expected minimum number hops random walk originating any gene reach outlier. HIT'nDRIVE thus aims find smallest set which one outliers within desired multi-hitting time. For it estimates time independent hitting times given outlier reduces weighted multi-set cover problem, solves integer linear program ILP. We apply data make phenotype predictions using only potential - more accurately than alternative approaches.