Accuracy of binding mode prediction with a cascadic stochastic tunneling method

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Accurate prediction of receptor-ligand binding modes is crucial for rational drug design.
• Existing computational methods face challenges in reproducing experimental binding poses.

Purpose of the Study:

• To evaluate the accuracy of the atomistic FlexScreen approach for predicting binding modes.
• To assess the performance of a forcefield-based scoring function in molecular docking.

Main Methods:

• Utilized the FlexScreen approach for atomistic simulations.
• Employed a simple forcefield-based scoring function.
• Tested on 83 complexes from the ASTEX/CCDC high-resolution database.

Main Results:

• Achieved a median RMS deviation of 0.83 Å between experimental and predicted binding modes.
• Over 80% of ligands docked within 2 Å of the experimental pose.
• The protocol successfully located the correct binding mode in 60 complexes across ten simulations.

Conclusions:

• The FlexScreen approach demonstrates high accuracy in predicting binding modes.
• Crystal water and clashes in experimental structures were identified as sources of docking failures.
• This method shows promise for applications in molecular docking and drug discovery.