Robust elastic network model: A general modeling for precise understanding of protein dynamics

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Normal mode analysis using elastic network models (ENMs) is widely used for studying protein dynamics.
• Standard ENMs often show low correlation with experimental B-factors due to neglecting crystal packing and using arbitrary spring constants.

Purpose of the Study:

• To develop a more accurate elastic network model for protein dynamics.
• To improve the correlation between simulated and experimental atomic fluctuations.

Main Methods:

• Proposed a robust elastic network model (RENM) incorporating crystalline effects via symmetric constraints.
• Utilized lumped masses and amino acid-specific spring constants based on chemical interactions.
• Validated the model on over 500 protein structures and 80 high-resolution crystal structures with anisotropic displacement parameters (ADPs).

Main Results:

• RENM significantly improves the correlation with experimental B-factors compared to standard ENMs.
• The model achieves this improvement without additional computational burden.
• An optimal spring constant was determined, enabling prediction of atomic fluctuations and vibrational density of states (VDOS) without fitting.
• RENM accurately describes vibrational characteristics of individual residues, including anisotropic displacement parameters.

Conclusions:

• RENM offers a more accurate and robust approach to simulating protein dynamics.
• The model provides quantitative predictions of atomic fluctuations and vibrational properties.
• RENM's ability to incorporate crystal packing and residue-specific properties enhances its applicability in structural biology.