A Local Rigid Body Framework for Global Optimization of Biomolecules

Area of Science:

• Computational chemistry
• Biomolecular simulations
• Molecular modeling

Background:

• Simulating biomolecules requires significant computational resources.
• Reducing degrees of freedom can enhance simulation efficiency.
• Current methods may not fully leverage rigid body approximations.

Purpose of the Study:

• To develop and evaluate a local rigid body framework for biomolecular simulations.
• To assess the computational speedup for global optimization using this framework.
• To determine the intrinsic rigidity of different molecular groups.

Main Methods:

• Implementing a local rigid body framework where atom sets are treated as rigid bodies.
• Utilizing a basin-hopping algorithm for global minimum searches.
• Benchmarking on tryptophan zipper (trpzip 1) and chignolin using CHARMM and AMBER force fields.

Main Results:

• Global minimum identification was 4.2x faster for trpzip 1 and 2.5x faster for chignolin.
• Rigidity order determined: side chain rings > termini > trigonal planar centers ≥ peptide bonds ≫ side chains.
• Framework shows potential for larger biomolecule structure prediction.

Conclusions:

• The local rigid body framework offers substantial computational savings in biomolecular simulations.
• The degree of rigidification impacts efficiency and accuracy.
• This approach is promising for advancing biomolecular structure prediction.