A robust algorithm for optimizing protein structures with NMR chemical shifts.

Area of Science:

• Structural biology
• Computational chemistry
• Biophysics

Background:

• Protein structure determination often relies on Nuclear Magnetic Resonance (NMR) data.
• Existing methods struggle to refine approximate protein models using NMR chemical shifts alone.
• There is a need for robust methods to improve protein structural models with limited experimental data.

Purpose of the Study:

• To introduce a novel computational method, CS-GAMDy, for refining protein structures.
• To demonstrate the efficacy of CS-GAMDy in optimizing protein models using NMR chemical shift data.
• To enable the generation of high-quality, PDB-worthy protein structures from sparse or approximate models.

Main Methods:

• Development of Chemical Shift driven Genetic Algorithm for biased Molecular Dynamics (CS-GAMDy).
• Integration of knowledge-based scoring functions and structural information from NMR chemical shifts.
• Utilizing multi-objective molecular dynamics (MD) biasing, a genetic algorithm, and XPLOR for protein modeling.

Main Results:

• CS-GAMDy successfully refines and folds protein models up to 10 Å (RMSD) from the correct structure using only NMR chemical shifts.
• The method refines a variety of approximate or erroneous protein structures towards known/correct structures and chemical shifts.
• Performance comparison shows CS-GAMDy is effective across various refinement scenarios.

Conclusions:

• CS-GAMDy offers a robust solution for refining protein structures using NMR chemical shift data.
• The method significantly improves the quality of protein models, making them suitable for deposition in the Protein Data Bank (PDB).
• CS-GAMDy is user-friendly, easily installed, and available for broader application in structural biology.