Computational Study of Estrogen Receptor-Alpha Antagonist with Three-Dimensional Quantitative Structure-Activity Relationship, Support Vector Regression, and Linear Regression Methods

Human estrogen receptor (ER) isoforms, ER alpha and ER beta, have long been an important focus in the field of biology. To better understand the structural features associated with the binding of ER alpha ligands to ER alpha and modulate their function, several QSAR models, including CoMFA, CoMSIA, SVR, and LR methods, have been employed to predict the inhibitory activity of 68 raloxifene derivatives. In the SVR and LR modeling, 11 descriptors were selected through feature ranking and sequential feature addition/deletion to generate equations to predict the inhibitory activity toward ER alpha Among four descriptors that constantly appear in various generated equations, two agree with CoMFA and CoMSIA steric fields and another two can be correlated to a calculated electrostatic potential of ER alpha