Comparative quantitative structure–activity relationship study of some 1-aminocyclopentyl-3-carboxyamides as CCR2 inhibitors using stepwise MLR, FA-MLR, and GA-PLS

摘要

Multiple linear regression (MLR), factor analysis in combination with multiple linear regression (FA-MLR), and genetic algorithm subset selection partial least square (GA-PLS) regression methods were used for quantitative structure–activity relationships (QSAR) model building. These approaches were employed to investigate the correlation between pIC₅₀ and various physicochemical descriptors of 28 compounds of 1-aminocyclopentyl-3-carboxyamides including substituted tetrahydropyran moieties as CCR2 inhibitors. The obtained models were validated using cross-validation and external test set. The predictability and robustness of the developed models were considered by some figures of merit such as RMSEP and Y-randomization. MLR, FA-MLR, and GA-PLS have R 2 equal to 0.84, 0.69, and 0.93, respectively. Predicted variance by MLR, FA-MLR, and GA-PLS (R 2 test) is 78, 75, and 78%, respectively. Furthermore, the domain of applicability which indicates the area of reliable predictions is defined. The prediction results by models are in good agreement with the experimental value.