Interplay of docking, pharmacophores, and shape in virtual high-throughput screening

CINF 93

Erik Evensen, Hans E. Purkey, Kenneth E. Lind, and Erin K. Bradley. Computational Sciences, Sunesis Pharmaceuticals Inc, 341 Oyster Point Blvd., South San Francisco, CA 94080
We have observed recently that post-filtering docking results using pharmacophore models leads to improved enrichments in virtual screening exercises over docking or pharmacophore screening alone. This counter-intuitive insight leads to further questions that point to potential areas for improvement in virtual high-throughput screening. For example, it has been proposed that the improvement over pure pharmacophore-based methods is because docking selects compounds that are shape complementary to the target. We will present inquiries into improving pharmacophore post-filtering and using shape filtering as a higher throughput surrogate for docking. We evaluate the interplay and impact of these methods by applying them to data sets obtained on multiple proteins from different families.