Comparing molecular dynamics force fields in the essential subspace.

Area of Science:

• Computational chemistry
• Biophysics
• Structural biology

Background:

• Molecular dynamics (MD) simulations are crucial for understanding protein behavior.
• Advancements in force fields and sampling methods enable longer, more accurate simulations.
• Previous studies showed varying agreement between MD simulations and experimental NMR data.

Purpose of the Study:

• To analyze and compare multi-microsecond MD simulations of ubiquitin and Protein G using different force fields.
• To investigate the relationship between force field parameterization, conformational ensembles, and experimental data agreement.
• To provide insights for selecting appropriate force fields for protein simulations.

Main Methods:

• Performed multi-microsecond molecular dynamics simulations of ubiquitin and Protein G.
• Utilized multiple distinct protein force fields with varying parameterization philosophies.
• Conducted structural and dynamic analyses of simulation trajectories.
• Compared simulation results with existing experimental NMR data.

Main Results:

• Coarse-grained motional properties were largely conserved across several force fields.
• Fine-grained structural and dynamic differences were observed between proteins and force fields.
• Force fields optimized for secondary structure balance were indistinguishable from others in these simulations.
• Long simulations (up to 10 microseconds) offered limited information for refining torsion parameters.

Conclusions:

• Force field choice significantly impacts detailed protein structure and dynamics in MD simulations.
• Comparison with experimental data is essential for validating and selecting force fields.
• Simulations of stable, folded proteins may not sufficiently inform secondary structure parameterization.
• Further development is needed to improve force fields for accurate secondary structure prediction.