Integrating Multiple Receptor Conformation Docking and Multi Dimensional QSAR for Enhancing Accuracy of Binding Affinity Prediction

Background: The accuracy of molecular conformation for Quantitative Structure Activity Relationship (QSAR) studies is an important criteria, and the most favourable bioactive conformer selection is a tough task. Correct ligand alignment as input for 3D-QSAR is an important step that is prone to human biases. Multiple-dimensional QSAR (mQSAR) approach provides a promising alternative to classic 3D-QSAR for drug discovery purposes. Objective: Obtaining ligand conformations from multiple receptor conformation docking (MRCD) will reduce the margin of error by incorporating the receptor based alignment of ligand conformations. To validate this assumption we performed 6D QSAR studies on reported HIV-1 protease inhibitors using Quasar 6.0. Materials & Method: The ensemble of conformation was obtained by MRCD of ligands in thirteen crystal structures of HIV-1 protease. 6D QSAR model was built using 65 cyclic urea molecules reported as HIV-1 protease inhibitors. Predictive ability of the model was validated using 35 cyclic urea molecules as test set. External predictive ability of the model was evaluated using a set of 24 HIV-1 protease inhibitors having varied structural scaffold. Result: 6D QSAR model obtained showed a reliable cross-validated r2(q2) of 0.899, r2(classic) of 0.908 and yielded a predictive r2 (p2) of 0.527. The ratio of q2/r2 was 0.991 and p2/q2 was 0.586 for external test set. Conclusion: The QSAR results invariably suggest that our approach is suitable for the identification of molecules having HIV-1 protease inhibitory potency. The underlying philosophy combines flexible docking for the identification of the binding modes and 6D QSAR for their quantification.