eHiTS: Docking and scoring ligand/target interactions to give good score-rmsd and ic50 correlations in in silico high throughput screening.

COMP 208

Danni Harris, danni@simbiosys.ca and Zsolt Zsoldos. Computational Chemistry, SimBioSys Inc, 135 Queen's Plate Dr, Suite 520, Toronto, ON M9W 6V1, Canada
Adequate treatment of complex interactions, such as pi-cation, non-conventional hydrogen bonding, pi-stacking in ligand -protein/receptor recognition can be crucial to accurate pose prediction in in silico virtual screening of compounds for diverse endpoints such as drug-discovery, predictive metabolism, and toxicity prediction. The ability of this docking/scoring approach to provide good score-based low RMSD discrimination of (correct) biochemically/pharmacologically relevant poses and IC50 correlations is shown with illustrations from several systems: nicotinic acetylcholine receptors and their surrogate binding proteins, kinases, and cytochrome P450s. The use of customized family training to improved pose prediction accuracy and lnKd/ln(IC50) correlations for specific user problems will be discussed as an method to improve discrimination of ligand-target recognition. A critical analysis is presented examining the merits of docking scoring functions as compared to quantum mechanical (QM) and molecular mechanics Poisson-Boltzmann free energy scoring of pharmacologically relevant poses.
 

Drug Discovery
8:30 AM-11:40 AM, Thursday, March 26, 2009 Salt Palace Convention Center -- 258, Oral

Division of Computers in Chemistry

The 237th ACS National Meeting, Salt Lake City, UT, March 22-26, 2009