Impact of multiple conformations on fingerprint descriptors uses in QSAR

COMP 230

Pierre-Jean L'Heureux, lheureup@iro.umontreal.ca1, Benoit Cromp, crompb@iro.umontreal.ca1, Shi Yi Yue, ShiYi.Yue@astrazeneca.com2, and Yoshua Bengio, bengioy@iro.umontreal.ca1. (1) Department IRO, Université de Montréal, 2920 Chemin de la tour, C.P. 6128, Succ. Centre-Ville, Montreal, QC H3C 3J7, Canada, (2) Department of Chemistry, AstraZeneca R&D Montreal, 7171 Frederick-Banting, St-Laurent, QC H4S 1Z9, Canada
Some recent studies on pharmacophore fingerprints have outlined the degredation of signal to noise ratio when using multiple conformations to extract the fingerprints. We will show examples of this degradation. In addition, we show the clustering of compounds in feature space as a mean to compare the power of different fingerprints to seperate biologically active and inactive compounds.