Accurate Binding Free Energy Method from End-State MD Simulations

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Estimating binding free energies (BFEs) is crucial for drug discovery.
• Current methods often require significant computational resources or lack accuracy.

Purpose of the Study:

• Introduce a novel computational strategy to estimate BFEs.
• Improve accuracy and efficiency in predicting protein-ligand interactions.

Main Methods:

• Utilized end-state molecular dynamics (MD) simulation trajectories.
• Integrated Linear Interaction Energy (LIE) with ANI-2x neural network potentials (machine learning).
• Employed the Atomic Simulation Environment (ASE) for atomic simulations.

Main Results:

• Achieved high accuracy (R = 0.87-0.88) correlating with experimental binding free energies.
• Demonstrated superior performance compared to existing end-state methods.
• Showcased reduced computational cost for BFE calculations.
• Enabled comparison of BFEs for ligands with diverse chemical scaffolds.

Conclusions:

• The proposed method provides an accurate and efficient approach for BFE estimation.
• This tool can accelerate the drug discovery process by enabling rapid screening of potential drug candidates.
• The open-source code facilitates broader adoption and further development in computational chemistry.