Ligand Based CoMFA, CoMSIA and HQSAR Analysis of CCR5 Antagonists
Changdev Gorakshnath Gadhe, Sung Haeng Lee, Thirumurthy Madhavan, Gugan Kothandan, DuBok Choi, Seung Joo Cho · Bulletin of the Korean Chemical Society · 2010
In this study, we have developed QSAR models for a series of 38 piperidine-4-carboxamide CCR5 antagonists using CoMFA, CoMSIA and HQSAR methods. Developed models showed good statistics in terms of $q^2$ and $r^2$ values. Best predictions obtained with standard CoMFA model ( $r^2$ = 0.888, $q^2$ = 0.651) and combined CoMSIA model ( $r^2$ = 0.892, $q^2$ = 0.665) with electrostatics and H-bond acceptor parameter. The validity of developed models was assessed by test set of 9 compounds, which showed good predictive correlation coefficient for CoMFA (0.804) and CoMSIA (0.844). Bootstrapped analysis showed statistically significant and robust CoMFA (0.968) and CoMSIA (0.936) models. Best HQSAR model was obtained with a $q^2$ of 0.662 and $r^2$ of 0.936 using atom, connection, hydrogen, donor and acceptor as parameters and fragment size (7-10) with optimum number of 6 components. Predictive power of developed HQSAR model was proved by test set and it was found to be 0.728.