Significant enhancement of docking sensitivity using implicit ligand sampling.

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Accurate quantification of protein-ligand interactions is crucial but challenging for computational methods.
• Current docking methods often fail due to inadequate handling of protein flexibility and complex dynamics.
• Existing scoring schemes neglect the dynamic nature of protein-ligand complexes.

Purpose of the Study:

• To develop a novel methodology, the 'ligand-model' concept, for enhanced sampling of protein conformations during ligand binding.
• To improve the accuracy of predicting binding free energies and native-like binding poses.
• To address limitations in current computational approaches for protein-ligand interaction studies.

Main Methods:

• The 'ligand-model' concept involves molecular-dynamics (MD) simulations with a virtual ligand that dynamically changes shape.
• This approach generates an ensemble of protein conformations relevant to diverse ligand binding.
• Docking is performed into representative structures from the MD simulation, followed by pose clustering and averaged score reranking.

Main Results:

• The ligand-model approach significantly improves the prediction of native-like binding poses.
• It enhances the quantification of binding affinities compared to traditional methods.
• Demonstrated superior performance over static docking and ensemble docking using apo MD simulations.

Conclusions:

• The ligand-model concept offers a more accurate and robust method for studying protein-ligand interactions.
• This methodology effectively incorporates protein flexibility and dynamics into binding predictions.
• The approach holds promise for advancing computational drug discovery and molecular modeling.