QSPR models for prediction of half wave potentials of some chlorinated organic compounds using SR-PLS and GA-PLS methods

The half-wave reduction potential is an important electrochemical property used for the characterization of organic compounds. This property, which is a characteristic constant for a reversible oxidation-reduction system, can be useful for predicting electrochemical properties of other organic compounds. In this work, quantitative structure-property relationship (QSPR) models have been introduced for estimating polarographic half-wave reduction potentials of 21 chlorinated organic compounds. Two QSPR models have been developed based on genetic algorithm-partial least squares (GA-PLS) and stepwise regression-partial least squares (SR-PLS) to predict half-wave potentials (E_1/2) of some chlorinated organic compounds. Variable selection is very important for QSPR modelling. In the present study, two selection variables methods were compared to choose molecular descriptors for the construction of a model by the PLS method. Both GA-PLS and SR-PLS methods resulted in accurate prediction, with more accurate results obtained by the GA-PLS model. The respective root mean square error of the prediction set obtained by the GA-PLS and SR-PLS models were 0.082 and 0.1302.