Grow to Fit Molecular Dynamics (G2FMD): an ab initio method for protein side-chain assignment and refinement

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Physics-based models for protein structure prediction face challenges due to rough energy landscapes and tight packing.
• Existing rotamer-based methods have limitations in accuracy and applicability.

Purpose of the Study:

• To develop an ab initio physics-based method for protein side-chain assignment and refinement.
• To overcome limitations of current models by addressing energy landscape roughness and steric hindrances.

Main Methods:

• Utilized molecular mechanics simulations with reduced side-chain size to create a smoother energy landscape.
• Employed molecular dynamics simulations for gradual side-chain regrowth and refinement based on interaction energies.
• Ensured resulting structures are free from steric collisions for subsequent all-atom refinement.

Main Results:

• Achieved nearly 100% accuracy for chi1 and chi2 of buried residues.
• 79% of predicted side-chain conformations were within 10 degrees of the native structure.
• Accuracy decreased for exposed side chains, indicating areas for further development.

Conclusions:

• The proposed ab initio method effectively predicts protein side-chain conformations, particularly for buried residues.
• The approach successfully mitigates issues of energy landscape roughness and steric packing.
• Further research is needed to improve accuracy for exposed residues and expand method applications.