COMP 178 |
| The size of molecular libraries steadily increases not only due to progress in combinatorial chemistry. Thus, molecular docking methods must cope with the rapidly increasing amount of compounds and several new application scenarios arise, e.g., finding appropriate scaffolds for new pockets, choosing suitable linkers to bridge the gap between two sub-pockets, or selecting focused sublibraries. Furthermore, the scoring problem still remains to be solved. Whilst not having a universal scoring function at hand, our approach to nonetheless come up with chemically sensible solutions strongly relies on filtering to rule out evidently wrong poses as early as possible - at best prior to docking. Thus, we integrated several filters before, during, and after docking. This includes receptor based pharmacophore constraints and extends to both standard and combinatorial screenings. While maintaining quality, we were able to speed up sequential docking up to a factor of 10 (i.e. ~1s/compound). Combining FlexX-Pharm with combinatorics even leads to much higher speed-up factors. Besides a priori target knowledge experience from runs on test data sets helps to define filters and gain new insights. Assisted by a newly developed environment for detailed interactive docking analysis, it shall be demonstrated how, for example, buriedness-based approaches or pharmacophores plus new scoring schemes derived by such analyses do improve the screening results for both standard and combinatorial libraries. |
|
Current Trends in Molecular Docking and Virtual Screening
1:00 PM-4:15 PM, Tuesday, 12 September 2006 Moscone Center -- Room 236, Oral
Division of Computers in Chemistry |