The automated optimisation of a coarse-grained force field using free energy data

Area of Science:

• Computational chemistry
• Molecular modeling
• Biophysics

Background:

• Atomistic models offer high detail but are limited for large, long-timescale simulations.
• Coarse-grained (CG) models accelerate simulations by reducing complexity, but often sacrifice accuracy.
• Improved parameterization methods are needed to enhance CG model quality and applicability.

Purpose of the Study:

• To develop an automated approach for optimizing coarse-grained force fields.
• To enhance the accuracy and applicability of CG models in molecular simulations.
• To reproduce free energy data from atomistic simulations for CG model parameterization.

Main Methods:

• Implemented an automated optimization process using ForceBalance.
• Utilized hydration free energy gradients as a target for force field optimization.
• Applied the method to optimize uncharged side-chains and protein backbones in the SIRAH CG force field.

Main Results:

• Successfully optimized parameters for the SIRAH protein coarse-grain force field.
• Optimized parameters closely reproduced hydration free energies from atomistic models.
• Achieved improved agreement with experimental data compared to previous models.

Conclusions:

• The automated approach effectively optimizes CG force fields using free energy data.
• This method enhances the accuracy of CG models, bridging the gap between atomistic detail and simulation speed.
• Optimized CG models show improved predictive power for molecular systems.