Improving accuracy in protein-ligand affinity calculations

COMP 104

Michael K. Gilson1, Chia en Chang2, and Wei Chen1. (1) Center for Advanced Research in Biotechnology, University of Maryland Biotechnology Institute, 9600 Gudelsky Drive, Rockville, MD 20850, (2) Department of Chemistry, University of Maryland, College Park, MD 20742
Although there is a need for greater accuracy in models of protein-ligand binding, it is not clear what direction to take to reach this goal. Analysis of relatively simple systems can provide valuable insight. We have studied 30 host-guest systems in aqueous and organic solvents, using a second-generation Mining Minima method. This method uses well-developed theory to provide the standard free energy of binding, along with information on conformational preferences and changes in configurational entropy. The computed affinities are accurate here to within about 1.5 kcal/mol. We observe large changes in configurational entropy that vary over multiple kcal/mole and do not correlate with simple descriptors. Some cases involve induced fit of the host to the guest. Such results suggest that computing protein-ligand binding affinities to “chemical accuracy” of ±1 kcal/mol will require moving to relatively detailed and precise models that include a careful treatment of configurational entropy.

Docking and Scoring
1:30 PM-4:50 PM, Tuesday, August 24, 2004 Pennsylvania Convention Center -- 109B, Oral

Division of Computers in Chemistry

The 228th ACS National Meeting, in Philadelphia, PA, August 22-26, 2004