Algorithms for detecting significantly mutated pathways in cancer

摘要: Recent genome sequencing studies have shown that the somatic mutations drive cancer development are distributed across a large number of genes This mutational heterogeneity complicates efforts to distinguish functional from sporadic, passenger Since hypothesized target relatively small cellular signaling and regulatory pathways, common approach is assess whether known pathways enriched for mutated However, restricting attention will not reveal novel or An alterative strategy examine in context genome-scale interaction networks include both well characterized additional gene interactions measured through various approaches We introduce computational framework de novo identification subnetworks network significant patients includes two major features First, we diffusion process on define local neighborhood “influence” each Second, derive two-stage multiple hypothesis test bound false discovery rate (FDR) associated with identified these algorithms human protein-protein using mutation data recent studies: glioblastoma samples The Cancer Genome Atlas lung adenocarcinoma Tumor Sequencing Project successfully recover be important cancers, such as p53 pathway also identify Notch pathway, been implicated other cancers but previously reported Our first, our knowledge, demonstrate computationally efficient statistically anticipate find increasing use increase size scope.