PROJECT 2: MAPPING THE PHARMACOGENETIC LANDSCAPE FOR PRECISION MEDICINE SUMMARY With the recent outpouring of somatic mutation data, we now have an extensive list of the genes involved in cancer. To realize the full potential of this information, we must now use technologies to functionally characterize the mechanisms of how these mutations are integrated to regulate and deregulate cellular networks. The premise of this project is that somatic mutations converge into genetic interaction networks, and these networks bring together mutations of all varieties, including genes with low frequency of oncogenic mutations, and tumor suppressor profiles. We hypothesize that by systematically mapping these networks and quantifying how specific mutations regulate these networks, new molecular targets for cancer therapy can be identified. The focus of this proposal is on invasive breast cancer (BC) and head and neck squamous cell carcinomas (HNSCC), diseases that together result in over 700,000 deaths each year worldwide. To this end, the aims of Project 2 focus on using stateoftheart highthroughput epistasis mapping and computational approaches to systematically interrogate the functions of individual genes and genepairs in breast and head and neck cancer pathology. Coupled with functional validations in preclinical models and clinical trial data, our approach will result in unprecedented insights into the underlying tumor biology as well as unraveling of the genetic vulnerabilities of direct therapeutic relevance. To realize this goal, the CCMI will leverage technologies and synergize expertise in the Krogan, Mali, Ideker, Grandis, Gutkind, Ashworth, Mesirov, van ‘t Veer and Esserman laboratories. The cornerstone of our approach will be utilization of CRISPRCas9 based reverse genetic screening methodologies, which our team has established, that enables de novo discovery of genetic interactions by targeting single or pairs of genes in a highthroughput fashion. Leveraging TCGA data, we propose to interrogate high value panels of the most frequently mutated, amplified, or deleted genes in BC and then separately do the same for HNSCC. High throughput screening of the order of 10,000 interactions per experiment will allow us to exhaustively map the underlying genetic interactions between these gene sets (Aim 1). Synthesizing this genetic interaction data with other repositories of genetic information, we will next build an integrative framework to identify highly conserved synthetic lethal interactions (Aim 2). We will validate this resource by testing drug sensitivity predictions in vitro as well as in existing pharmacogenomic data sets. Finally, to facilitate clinical translation of the highestconfidence genegene and genedrug interactions, we will utilize our ...