Potential functions for virtual screening and ligand binding calculations: Some theoretical considerations

COMP 105

Kim A Sharp, Biochemistry and Biophysics, Biochemistry and Biophysics, University of Pennsylvania, 3700 Hamilton Walk, Philadelphia, PA 19104
Virtual screening requires identifying rare candidates with significant affinity for a known target structure. This requires identification of the 'best' ligand conformation/position/orientation (pose), ie. determination of the binding affinity of a given pose (at least relative to another pose): the 'binding problem'. Any method that successfully scores candidates/poses must emulate, however approximately, the true binding free energy. I will discuss: the statistical mechanics of binding, the requirements for a true free energy calculation, and some practical principles for evaluating and improving screening potentials. This includes determining whether a screening potential is deficient because the potential is unphysical, has an inadequate functional form, or because the parameters need to be improved. In the first two cases, no amount of parameterization with training/test sets will improve the screening potential significantly. Effort would be better spent re-tooling the potential function. Some issues specific to particular types of scoring/binding potentials will be discussed.
 

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