Three-dimensional vector of pharmacophore atomic interaction and its applications to stereoisomeric description

A novel 3D-molecular structure characterization method, three-dimensional vector of pharmacophore atomic interaction (3D-Vpai), saving 3D-structural information of compounds by the factor of stereo-sensitive angle (FSSA) was defined to identify stereoisomers. The predictions of 31 steroids benchmarks and 79 carbapenem derivatives using multiple linear regression analysis were carried out. The statistically significant QSAR models were obtained (steroids benchmarks: N = 25, R2 = 0.894,; Rcv2 = 0.785; carbapenem derivatives: n = 64, R2 = 0.821, Rcv2 = 0.765). Finally, classification of 32 stereoisomers of perindoprilate as angiotensin-converting enzyme inhibitor using linear discriminant analysis (LDA) was carried out. All LDA models classified correctly 100% for active and inactive compounds. Our results showed 3D-Vpai models remarkably superior to others, and it provides a powerful alternative to other more common 3D-QSAR descriptors.