Multi-body effects in a coarse-grained protein force field.

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Coarse-grained (CG) models simplify complex biomolecular systems by reducing degrees of freedom.
• CG models enable simulations at longer time and length scales than all-atom models.
• Emergent multi-body interactions are expected in formally derived CG models and improve accuracy.

Purpose of the Study:

• To develop a systematic approach for constructing CG effective energies with arbitrary orders of multi-body terms.
• To utilize a neural network-based method for building CG models as multi-body expansions.
• To evaluate the contribution of different multi-body terms to model accuracy.

Main Methods:

• A novel neural network-based approach was developed to construct CG models.
• The method constructs CG models as a systematic multi-body expansion.
• The approach was applied to a small protein system for validation.

Main Results:

• The study demonstrates a neural network-based strategy for CG model construction.
• The multi-body expansion for CG models shows slow convergence.
• Accurate reproduction of atomistic free energy requires up to five-body interactions in the CG model.

Conclusions:

• The proposed neural network approach provides a systematic way to build CG models.
• Higher-order multi-body interactions are essential for accurate CG model representation.
• Understanding multi-body interaction importance guides the development of more predictive CG models.