Postprocessing of docked protein-ligand complexes using implicit solvation models.

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Molecular docking is crucial for structure-based drug design, aiding in identifying bioactive conformations and ranking ligand binding affinities.
• Accurate prediction of protein-ligand interactions is essential for effective drug discovery pipelines.

Purpose of the Study:

• To investigate the impact of implicit water models on postprocessing molecular docking-generated binding poses.
• To enhance the accuracy of binding pose prediction and the correlation with experimental binding data.

Main Methods:

• Utilized molecular mechanics Poisson-Boltzmann/generalized Born surface area (MM-PB/GB-SA) methodology for postprocessing.
• Performed geometry optimization of binding poses using GB solvation models and energy minimization of the binding site.
• Calculated free energies of binding using the PB implicit solvent model and rescored poses.

Main Results:

• Geometry optimization improved binding pose accuracy for 20% of investigated complexes.
• Minimizing the binding site energy with GB models significantly reduced computational time compared to minimizing the entire complex with PB.
• Optimizing poses with GB(HCT+SA) and calculating binding free energies with PB yielded poses similar to crystal structures.
• Rescoring top-ranked poses improved correlations with experimental binding data.

Conclusions:

• The developed postprocessing protocol, incorporating implicit water models, offers a generally useful approach for structure-based drug discovery.
• Applying postprocessing to multiple top-ranked poses, rather than just the top one, further improves predictive accuracy.
• The protocol was successfully validated on Factor Xa inhibitors and MHC A(q) protein ligands.