Using AM1 Hamiltonian and factor analysis in prediction of partition properties for phenylthio, phenylsulfinyl and phenylsulfonyl acetates

Basing on the descriptors computed by AM1 hamiltonian for thirty phenylthio, phenylsulfinyl and phenylsulfonyl, acetates, the application of factor analysis and multiple regression analysis, we obtained two equations, which can be used to predict octanol/water coefficients and HPLC capacity factors for similar sulfur-containing compounds. Factor analysis overcomes the problem caused by intercorrelated date sets and allows the extraction of information useful for the determination of physico-chemical properties. Tbe present study showed the feasibility of using factor scores as descriptors in QSPRs. Analyzing the loadings of principal factors by individual structural descriptors, one can obtain the information condensed in each factor. This study provides a new analysis method for QSPR studies, which permits near a priori prediction of partition properties for compounds for which the empirical parameters are not readily available. Copyright (C) 1996 Elsevier Science Ltd