Title:

Enrichment enhancement in high throughput virtual screening through a
combination of FRED, ROCS, EON scores and naïve Bayesian classification.

Authors:

John E. Knox and Lakshmi S. Narasimhan
Pfizer

Abstract:

High throughput docking and scoring using FRED and removal of potential
false-positives through the use of a surface complementarity based scoring
function (Pfiscore) gives acceptable enrichment of actives in the top 0.5
to 5% of databases. We have studied the level of enrichment achieved when
ROCS and EON scores are also used and consensus scoring is used to rank
compounds. In addition, we have investigated the application naïve
Bayesian classifier to identify actives using FRED, ROCS and EON scores in
addition to fingerprints. The addition of fingerprints to the scores
complements the scoring and provides significant enhancement of enrichment
of actives in the top 0.5 of the compound collection studied on Bayesian
classification and reranking. In this presentation we will discuss the
performance of this virtual screening workflow to 18 different test
datasets (actives distributed amongst decoys) and briefly cover the
effectiveness of this workflow in application to projects through
retrospective studies