Prediction of pesticides chromatographic lipophilicity from the computational molecular descriptors

Abstract

Quantitative structure-property relationship models were developed for the prediction of pesticides and some PAH compounds lipophilicity based on a wide set of computational molecular descriptors and a set of experimental chromatographic data. The chromatographic lipophilicity of pesticides has been evaluated by high-performance liquid chromatography (HPLC) using different chemically bonded (C18, C8, CN and Phenyl HPLC columns) stationary phases and two different organic modifiers (methanol and acetonitrile, respectively) in the mobile phase composition. Through a systematic study, by using the classic multivariate analysis, several quantitative structure-property/lipophilicity multi-dimensional models were established. Multiple linear regression and genetic algorithm for the variable subset selection were used. The internal and external statistical evaluation procedures revealed some appropriate models for the chromatographic lipophilicity prediction of pesticides. Moreover, the statistical parameters of regression and those obtained by applying t-test for the intercept (a(0)) and for the slope (a(1)) in order to evaluate relationship between experimental and predicted octanol-water partition coefficients in case of the test set compounds, revealed two statistically valid models that can be successfully used in lipophilicity prediction of pesticides.