...................................................................... Mathematical Modeling of Drugs to Treat Cocaine Abuse
Department of Chemistry & Environmental Science
Analysis of large, flexible, drug-like molecules is complicated by the fact that these molecules can take on a wide range of closely-related conformations. The first step in the analysis of such a molecule is to classify the conformers into groups. This cannot be done by inspection since the presence of multiple rotatable bonds produces a "continuum" of conformations, rather than a discrete set. We explore several approaches to classification, such as hierarchical and fuzzy clustering, and singular value decomposition. We compare these techniques in classification of a dataset of over 700 molecular conformations. We then use representative conformations taken from each of these groups in pharmacophore modeling to elucidate the molecular properties required of a drug useful in the treatment of cocaine abuse. Last Modified: Jan 18, 2006 Victor Matveev m a t v e e v @ n j i t . e d u |