فهرست نشریات علمی دسترسی آزاد ایران

The aim of this study is to investigate QSAR modeling for a set of novel pyridine derivatives as inhibitors of p38α enzyme. After calculating a set of molecular descriptors, the selection of variables was performed using two methods: genetic algorithm (GA) and stepwise regression (SW). To build and evaluate the models, the dataset was divided into two training and test sets based on the clustering method, which included 35 compounds in the training set and 10 compounds in the test set. The results showed that the genetic algorithm-based model (GA-MLR) performed better than the stepwise regression model (SW-MLR). The final GA-MLR model was developed using six descriptors and its statistical values were obtained as R²train = 0.835, RMSEtrain = 0.385, Ftrain = 26.199, R²test = 0.645, RMSEtest = 0.601 and Ftest = 1.065. In order to confirm the accuracy of the model, in addition to validation by the test set, cross-validation techniques, domain determination, and Y-randomization test were also performed. These results indicate that the developed model can be used as an effective tool for designing new pyridine derivatives with higher inhibitory potency and predicting their activity before synthesis.