1. Optimization of virtual screening protocols: FlexX based virtual screening for COX-2 inhibitors reveals the importance of tailoring screen parameters
- Author
-
Zoltán Kovári, György M. Keserű, and András Bauer
- Subjects
Virtual screening ,business.industry ,Chemistry ,Drug design ,Nanotechnology ,Condensed Matter Physics ,Machine learning ,computer.software_genre ,Biochemistry ,Docking (molecular) ,Artificial intelligence ,Physical and Theoretical Chemistry ,business ,computer - Abstract
As molecular docking poses a challenging problem in rational drug design many workgroups have lavished their attention on finding and testing various docking techniques in order to maximize their benefits. One such experiment we report was conducted using the FlexX docking algorithm on a variety of data sets containing COX-2 inhibitors as well as randomly selected inactive molecules which were docked into the COX-2 crystal structure to test the predictive power of scoring functions available in either FlexX or CScore. Our goal was to determine both their abilities to separate active and inactive molecules, as well as to find which function shows the best correlation with the molecules' IC 50 values. We have shown, that F-score had the best ability to rank the COX-2 inhibitors according to their IC 50 values, while ChemScore has proven itself to be the most adequate for finding possible hits in large databases.
- Published
- 2004
- Full Text
- View/download PDF