Thursday 6th June, 2013
3:00pm to 3:30pm
One of the long-standing methods for validation of a docking engine/scoring function is self- or cognate docking; re-docking a crystallographic ligand into its cognate protein receptor. Several problems have made determining if new scoring functions/docking engines do improve cognate docking; commonly used measures of deviation from the "correct" crystallographic pose are insufficiently powerful, datasets of protein-ligand complexes have not been assembled with appropriate care and commonly used statistical methods for determining difference and/or superiority for a given technique have difficulty with the highly non-normal character of data from these studies. We present a comparison of a variety of docking/scoring methods found in our docking engine OEDocking on our recently released dataset, Iridium. We apply rigorous statistical analyses to the data to determine if, and with what likelihood, a given technique outperforms another. With a well validated approach to cognate docking available we expect the field will be able to turn to the problem of genuine interest, cross-docking, with greater confidence than hitherto.
Sign in to add slides, notes or videos to this session