Informatics methods for understanding drug action

About Prof Altman’s lab:
Our lab has created a computational representation of protein microenvironments that captures their key biochemical and biophysical features. We have used this representation to build several capabilities that are useful for understanding drug action. These capabilities include (1) recognizing the similarities between protein pockets, (2) predicting the likelihood that a pocket specifically will bind a small molecule drug, (3) predicting small molecule fragments that may bind parts of a protein pocket, (4) predicting the side effects of a drug that binds a pocket based on its anticipated pattern of off-target binding, and (5) predicting the dose range for a drug that binds a pocket based on the expected tradeoff between promiscuity and selectivity. We are interested in using these tools for drug target identification and repurposing. If time allows, I will also discuss our work extracting gene-drug-disease associations from text using natural language processing techniques.