• 3D-QSAR;
  • Bayesian model;
  • benzodiazepine;
  • Pharmacophore;

Ligands binding at the benzodiazepine site of GABAA receptor play important pharmacological roles in clinical application. In this study, ligand-based pharmacophore modeling, 3D-QSAR analysis, and Bayesian model studies have been performed on a set of 84 diverse ligands binding at the benzodiazepine site. The results showed the best pharmacophore hypothesis AADHR.4, which included two hydrogen acceptors (A), one hydrogen donor (D), one hydrophobic group (H), and one aromatic ring (R). Atom-based 3D-QSAR model was built, and it showed good statistical significance (R2 = 0.936) and excellent predictive ability (Q2 = 0.821). Moreover, Bayesian model was developed and used to identify the key molecular features which are good or bad for the ligand binding activity. All the results from the pharmacophore, 3D-QSAR, and Bayesian modeling studies revealed that a hydrogen-bond donor (e.g., N-H) and a hydrophobic group (e.g., Br) are critical structural features for the ligands binding at the benzodiazepine site.