Compound efficacy, compound specificity, and target identification are crucial for successful cancer therapeutics. The accurate characterization of genetic determinants of compound sensitivity in cancer cells would advance drug discovery, particularly in patients with known genetic lesions, and ultimately improve outcomes for patients. We characterized NCI-60 cancer cell line compound-sensitivity profiles with respect to gene-expression measurements. We used a novel analysis method that combines a Spearman rank-correlation algorithm for feature selection with elastic net regression analysis for predictive model generation. Using this approach, we identified several candidate gene lists (weighted by importance) that appear to predict sensitivity. Using biological pathway analysis, we studied the underlying biological networks relevant to compound sensitivity and cancer. These biological networks suggest several targets for anti-cancer compound development, and provide a framework for continued analysis of the connection between compound efficacy/specificity analyses and genetic characterization of cancer cells.
PROJECT: Development of Predictive Models to Identify Pathways Relevant to Compound Sensitivity in Cancer
The Broad Institute is a model for what effective scientific communication and collaboration can achieve. The Broad is interdisciplinary at its core, and because of this, the pursuit of knowledge at the Broad is without bounds. All of us who have been fortunate enough to become a part of the Broad community know this, and have grown as scientists each day at the Broad. The future is in interdisciplinary research, and collaborative efforts to solve biomedical problems, and this where I look to apply myself; my experience here at the Broad has been inspiring, encouraging, and nothing short of amazing.