Development of Quantitative Structure−Binding Affinity Relationship Models Based on Novel Geometrical Chemical Descriptors of the Protein−Ligand Interfaces

• DOI: 10.1021/jm050260x
• Publication Date: 2006-04-27T04:36:55Z
• Authors: Shuxing Zhang; Alexander Golbraikh; Alexander Tropsha
• Keywords: Electronegativity; Protein ligand; Molecular descriptor

Novel geometrical chemical descriptors have been derived on the basis of computational geometry protein-ligand interfaces and Pauling atomic electronegativities (EN). Delaunay tessellation has applied to a diverse set 517 X-ray characterized complexes yielding unique collection interfacial nearest neighbor quadruplets for each complex. Each quadruplet composition was by single descriptor calculated as sum EN values four participating atom types. We termed these simple generated from with Tessellation ENTess used them in variable selection k-nearest quantitative structure-binding affinity relationship (QSBR) studies 264 known binding constants. Twenty-four chemically dissimilar ligands were aside an independent validation set, remaining dataset 240 divided into multiple training test sets. The best models leave-one-out cross-validated correlation coefficient q(2) high 0.66 R(2) 0.83 set. predictive power confirmed independently applying 24 0.85. conclude that QSBR built can be instrumental predicting receptor-ligand complexes.