Fast and accurate coarse-grained estimate of small molecule binding free energies

CINF 89

Jun Shimada, Alexey V. Ishchenko, Kam Jim, David J. Lawson, Peter R. Lindblom, Guosheng Wu, and JP Wery. Computational Drug Discovery, Concurrent Pharmaceuticals, Inc, 502 West Office Center Drive, Fort Washington, PA 19034
A novel approach for the prediction of binding free energies will be presented. This approach is characterized by three critical features: (1) a coarse-grained physical model of the binding process, (2) trainability, and (3) a sophisticated machine learning algorithm that maximally utilizes the information from bioassays. When used against multiple pharmacologically relevant targets, this scoring function has proven to be accurate and generalizable outside of the training set. In virtual screening against aspartyl proteases, nuclear receptors and kinases, this approach was able to select inhibitors which, after synthesis, were shown to be active.