Fast docking using the CHARMM force field with EADock DSS.

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Accurate prediction of binding modes (BMs) is crucial for drug development.
• Existing docking software varies in speed and accuracy, with some accurate methods being computationally expensive.
• EADock2 offers high accuracy but suffers from significant computational cost.

Purpose of the Study:

• To develop a faster molecular docking approach based on EADock2.
• To reduce the computational cost of accurate binding mode prediction.
• To maintain or improve the accuracy of docking predictions for drug-like molecules.

Main Methods:

• EADock dihedral space sampling (DSS) utilizes EADock2's hybrid sampling and multiobjective scoring.
• Incorporates an automatic bias for sampling putative binding sites.
• Employs an efficient tree-based DSS algorithm.

Main Results:

• EADock DSS achieves performance equivalent to EADock2 for drug-like ligands.
• CPU time is reduced by several orders of magnitude compared to EADock2.
• 57% of binding modes reproduced within 2 Å RMSD (top prediction); 70% within top five predictions.
• Cross-docking success rates are comparable to AutoDock with protein flexibility.

Conclusions:

• EADock DSS offers a significant computational speedup for molecular docking.
• The method maintains high accuracy for predicting binding modes of drug-like ligands.
• This approach enhances the efficiency of drug discovery pipelines.