Using diverse computational approaches to understand protein/ligand binding affinities in structure-based drug design: A cross-docking study

COMP 41

Johannes H. Voigt, johannes.voigt@spcorp.com1, Carl Elkin2, and Jose S. Duca1. (1) Structural Chemistry, Schering-Plough Research Institute, 2015 Galloping Hill Rd, Kenilworth, NJ 07033, (2) Schering-Plough Research Institute Cambridge, 320 Bent St, Cambridge, MA 02141
Predicting protein/ligand binding affinity is one of the most challenging computational chemistry tasks. Numerous methods have been developed to address this challenge, but they all have limitations. Addressing protein flexibility has been a shortcoming of many methods. In this study, we used cross-docking of ~150 inhibitors into the full set of crystal structures for each inhibitor complexed with the kinase CDK2. In scoring relative binding potency based on multiple combinations of several target proteins, the dangers of over-fitting became apparent. The performance of the GOLD and GLIDE docking programs was compared in this exhaustive study. The estimation of protein/ligand binding affinities from QM-MM calculations will also be discussed. Finally various methods of computing binding affinities will be assessed vis-à-vis their speed and accuracy.
 

Drug Discovery
1:00 PM-5:05 PM, Sunday, August 19, 2007 BCEC -- 161, Oral

Division of Computers in Chemistry

The 234th ACS National Meeting, Boston, MA, August 19-23, 2007