Comparative Data Analysis of Virtual Screening Methodologies for Predicting Urease Inhibitory Activity.

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and development
• Molecular modeling and simulation

Background:

• Structure-based virtual screening (SBVS) is crucial for identifying drug candidates.
• SBVS accuracy depends heavily on chosen methodologies, scoring functions, and data processing.
• Optimizing SBVS protocols is essential for efficient drug discovery pipelines.

Purpose of the Study:

• To systematically evaluate and compare different SBVS protocol variants for drug discovery.
• To assess the impact of data fusion techniques and pose number on ligand ranking accuracy.
• To refine SBVS workflows for enhanced ligand prioritization against therapeutic targets.

Main Methods:

• Employed five SBVS variants: molecular docking, induced-fit docking (IFD), quantum-polarized ligand docking (QPLD), ensemble docking (ED), and MM-GBSA.
• Utilized four crystallographic structures of *Helicobacter pylori* urease from the PDB.
• Assessed performance using correlation metrics (Spearman, Pearson) and error measures (MAE, RMSE, IRM), alongside six data fusion techniques.

Main Results:

• Ensemble docking (ED) and MM-GBSA demonstrated superior performance in compound ranking.
• MM-GBSA showed higher errors in absolute binding energy predictions compared to ED.
• The minimum data fusion approach proved robust across varying numbers of docking poses.
• pIC50 values yielded higher Pearson correlations for affinity prediction than IC50 values.

Conclusions:

• MM-GBSA and ED are effective methods for compound ranking in SBVS.
• Minimum data fusion is a reliable strategy for aggregating docking results.
• pIC50 is a more suitable metric for affinity prediction in correlation analyses.
• Optimized SBVS workflows enhance ligand prioritization for drug discovery campaigns.