Enhanced Exploration of Protein Conformational Space through Integration of Ultra-Coarse-Grained Models to Multiscale...

Area of Science:

• Biophysics
• Computational Biology
• Molecular Modeling

Background:

• All-atom (AA) molecular dynamics (MD) simulations offer detailed insights but are computationally intensive.
• Coarse-grained (CG) models simplify molecular representations to study larger systems and longer timescales.
• Ultra-coarse-grained (UCG) models further reduce complexity for enhanced computational efficiency.

Purpose of the Study:

• Develop and integrate UCG models into the Multiscale Machine-Learned Modeling Infrastructure (MuMMI).
• Enable efficient sampling of protein conformations and dynamics.
• Investigate protein-membrane interactions and conformational changes.

Main Methods:

• Developed UCG models using essential dynamics coarse graining (EDCG) and heterogeneous elastic network modeling (hENM) with anharmonic modifications.
• Utilized fluctuations from higher-resolution Martini CG simulations to parameterize UCG models.
• Incorporated an implicit membrane model for protein-membrane dynamics.
• Developed a machine-learning-based backmapping approach for UCG to Martini CG conversion.

Main Results:

• Achieved accurate sampling of protein configurations and long-range conformational changes.
• Enhanced exploration of protein-membrane dynamics through the implicit membrane model.
• Improved prediction accuracy using the novel machine-learning backmapping method.
• Demonstrated efficient conformational sampling of the RAS-RBDCRD protein complex.

Conclusions:

• UCG models integrated into MuMMI significantly advance the exploration of protein configurations.
• This approach offers critical insights into the role of protein dynamics in biological processes.
• The developed methods provide a powerful tool for large-scale molecular simulations.