Large scale relative protein ligand binding affinities using non-equilibrium alchemy

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Alchemical free energy calculations using molecular dynamics (MD) simulations are powerful for drug design.
• Complex setup procedures hinder the widespread adoption of these methods.
• Limited end-to-end open-source tools are available.

Purpose of the Study:

• To present an accessible, open-source approach for alchemical free energy calculations.
• To enable easy setup and execution of calculations for diverse small molecules.
• To validate the method's performance against established commercial tools.

Main Methods:

• Utilized the open-source software pmx with the GROMACS MD engine.
• Employed non-equilibrium thermodynamic integration (TI) for calculations.
• Combined results from Amber and Charmm force fields for consensus outcomes.

Main Results:

• Achieved high precision and accuracy in relative protein-ligand alchemical free energy calculations.
• Demonstrated performance on par with the commercial FEP+ approach.
• Obtained an average unsigned error (AUE) of 3.64 ± 0.14 kJ mol⁻¹ on a large dataset of 482 perturbations.

Conclusions:

• The presented pmx-based method offers a user-friendly, open-source alternative for alchemical free energy calculations.
• This approach facilitates broader application of accurate binding affinity predictions in drug design.
• The consensus results from multiple force fields enhance the reliability of the predictions.