Prediction of Activities of BRAF (V600E) Inhibitors by SW-MLR and GA-MLR Methods

Background: Quantitative structure–activity relationship (QSAR) models could provide both statistical significance and useful chemical insights for drug design. The QSAR method has found applications for predicting diverse properties of organic compounds, including antiviral activities, toxicities and biological activities. In this work, a quantitative structure-activity relationship was utilized for the prediction of allosteric BRAF (V600E) inhibitory activities. Methods: A data set which contains 54 molecules was classified into training and test sets. Stepwise (SW) and genetic algorithm (GA) methods were employed for feature selection. The models were validated using the cross-validation and external test set. Results: Results showed that the GA approach is a more powerful technique than SW for the selection of suitable descriptors. The squared cross-validated correlation coefficient for leave-one-out of 0.702 and squared correlation coefficient of 0.793 was obtained for the training set compounds by GA–MLR model. Conclusion: The obtained GA–MLR model could be applied as a worthwhile model for designing similar groups of the mentioned inhibitors.