Prediction of toxicity and octanol–water partition coefficient of Carbamate Derivativesas Insecticides Using Genetic Algorithm-Multiple Linear Regressions Method

A Quantitative Structure–Activity Relationship (QSAR) study based on Genetic Algorithm  Multiple Linear Regressions (GA-MLR) were carried out for the prediction of the toxicity (logIC50) and the logarithm of octanol-water partition coefficient (logPow) of some carbamate derivatives as insecticides. The optimized conformation of compounds were obtained at HF/6-31G* level with Gaussian 98 software. Dragon software is used to calculate molecular descriptors. A data set of these compounds was randomly divided into 2 groups: training and test sets. The QSAR models were optimized using multiple linear regressions (MLR).The most relevant molecular descriptors were collected by Genetic Algorithm (GA) and backward regression. The best GA-MLR models are obtained using statistical parameters, such as squared correlation coefficient (R2), adjusted squared correlation coefficient (R2adj), root mean square error (RMSE) values for training and test sets. The best QSAR models are obtained based on the statistical parameters Leave-one-out (LOO) cross-validation, external test set, external validation parameters (Q2F1, Q2F2, Q2F3) and the concordance correlation coefficient (CCC) were used to quantify the predictive ability of GA-MLR models. The results showed that GA-MLR models could be used to predict the activities of carbamate derivatives.