PRODIGY: personalized prioritization of driver genes

• DOI: 10.1101/456723
• Publication Date: 2018-10-31T00:06:28Z
• Authors: Gal Dinstag; Ron Shamir
• Keywords: Identification; Prioritization

Abstract Background Evolution of cancer is driven by few somatic mutations that disrupt cellular processes, causing abnormal proliferation and tumor development, while most have no impact on progression. Distinguishing those mutated genes drive tumorigenesis in a patient primary goal therapy: Knowledge these the pathways which they operate can illuminate disease mechanisms indicate potential therapies drug targets. Current research focuses mainly cohort-level driver gene identification, but patient-specific identification remains challenge. Methods We developed new algorithm for ranking genes. The algorithm, called PRODIGY, analyzes expression mutation profiles along with data known protein-protein interactions. Prodigy quantifies each every deregulated pathway using prize collecting Steiner tree model. Mutated are ranked their aggregated all pathways. Results In testing five TCGA cohorts spanning >2500 patients comparison to validated genes, outperformed extant methods based network centrality measures. Our results pinpoint pleiotropic effect show capable identifying even very rare drivers. Hence, takes step further towards personalized medicine treatment. Availability R package available at: https://github.com/Shamir-Lab/PRODIGY .