Receptor-based QSAR studies of non-peptide human oxytocin receptor antagonists.
In the present study, QSAR calculations were performed on the receptor-based alignment of 58 non-peptide human oxytocin receptor antagonists. With the aid of different scoring functions (AutoDock 3.05 built-in and X-Score 1.2) the evolved receptor-ligand complexes were characterized. By means of various datasets it was confirmed that the scoring functions were not capable to predict the biological activity correctly in compounds containing a rigid derivative in the variable region. To improve the pKi prediction 3D-QSAR calculation was performed. The regions related to the biological activity were determined by using cross-validated r2(q2)-guided region selection (q2-GRS) method. The predictive power of the CoMFA model [r(pred)2=0.89, q2(LMO, five groups)=0.695+/-0.034] allowed prediction of the biological activities of newly synthesized compounds and confirmed the receptor-based alignment.