COMP 368 |
| In recent years, several methods have been developed to calculate absolute binding affinities of protein-ligand complexes. Endpoint-based approaches, such as MM-PBSA, are computationally efficient, but are subject to significant errors in entropy estimation and require taking differences between large energetic values. On the other hand, path-based approaches intrinsically include a rigorous determination of entropy but are typically more computationally intensive. To reduce errors associated with energy differences, we employed a method whereby a ligand is spatially pulled out of the binding pocket of the protein and a potential of mean force is calculated. Umbrella sampling was used to hold the ligand at different points along the pulling path. Weighted histogram analysis was used to combine the data from multiple simulation windows. We compared results from explicit and implicit solvent simulations of two diverse FK506-binding protein (FKBP) complexes to experimentally observed binding affinities. In addition, we applied our approach to drug design, by calculating the binding affinities of candidate molecular inhibitors to protein targets of interest in our lab. |
|
Free Energy Computations in Drug Discovery
9:00 AM-12:40 PM, Thursday, 14 September 2006 Moscone Center -- Room 220/222, Oral
Division of Computers in Chemistry |