Predicting the Binding Affinity of ERβ Ligands Based on a Novel Variable Selection Method

A number of descriptors were employed to characterize the molecular structures of the 128 estrogen receptor β ligands. A quantitative structure-activity relationship (QSAR) model of these compounds was developed by the variable selection method based on variable interaction. The QSAR model with five descriptors was internally and externally validated. The determination coefficient (R ²) and the leave-one-out cross-validated correlation coefficient (Q ²) are 0.8272 and 0.8041, respectively. The estimated correlation coefficient of the external validation is 0.8255. The mechanistic interpretation of the final model was carried out according to the definition of descriptors. As the model meets the five principles proposed by Organization for Economic Co-operation and Development, it can be used to predict the binding affinity of other derivatives.