Comparing computational approaches to screening library selection


Erik Evensen, Hans Purkey, Ken Lind, and Erin K. Bradley. Computational Sciences, Sunesis Pharmaceuticals Inc, 341 Oyster Point Blvd., South San Francisco, CA 94080
Screening compound collections has long been a method for finding starting points in the drug discovery process. Modern automation techniques make it possible to screen increasingly large numbers of compounds. Nevertheless, the numbers of compounds and size of chemical space represented in corporate collections or accessible via convenient chemistries involving commercially available compounds exceeds the cost time, cost, and practical limitations of even the largest high throughput screening effort. Moreover, such screening is largely unproductive. It would be useful, therefore, to have a method for prioritizing or triaging compounds for acquisition and/or screening. We present a comparison of a number of computational approaches to such triage, including docking with scoring functions of varying complexity, searching for pharmacophore complementarity to the target, and prioritizing compounds to reveal shape or feature preferences of the site.