The Information Content of 2D and 3D Structural Descriptors Relevant to Ligand-Receptor Binding

作者：Robert D. Brown, Yvonne C. Martin · 发表于：Journal of Chemical Information and Computer Sciences · 年份：1997 · DOI：10.1021/ci960373c · 被引用次数：385 · 研究领域：Computational Drug Discovery Methods、Various Chemistry Research Topics、Chemical Synthesis and Analysis

We have previously studied the ability of various structural descriptors to distinguish between biologically active and inactive compounds (ref 1). This paper examines the degree to which these descriptors encode information relevant to the forces of ligand-receptor binding, namely hydrophobic, dispersion, electrostatic, steric, and hydrogen bonding interactions. This is assessed by the ability to accurately predict values for physical properties of a structure related to each of the interactions, from the known values for other structures which are shown to be structurally similar to the first by the descriptor in question. Our results suggest that the differences we observed in the ability of descriptors to separate active from inactive molecules may be explained by the degree to which they encode information relevant to ligand-receptor binding. In particular we found that the MACCS structural key descriptor implicitly contains a great deal of information relevant to each type of interaction.