A scoring function based on solvation thermodynamics for protein structure prediction.

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Protein structure prediction is crucial for understanding protein function.
• Accurate prediction requires robust energy functions and efficient sampling methods.
• Existing methods face challenges in achieving high accuracy for novel protein structures.

Purpose of the Study:

• To evaluate a new free energy function for protein stability, focusing on solvation thermodynamics.
• To assess the performance of this function when combined with fragment assembly (FA) and comparative modeling (CM) sampling methods.
• To identify limitations and potential improvements in protein structure prediction methodologies.

Main Methods:

• Development of a novel free energy function based on solvation thermodynamics.
• Integration of the energy function with fragment assembly (FA) and comparative modeling (CM) for structure prediction.
• Validation using 11 small proteins with known high-resolution crystal structures.
• Application of replica exchange molecular dynamics and coarse-grained normal mode analysis for refinement and analysis.

Main Results:

• Accurate protein models (average Cα RMSD ~2.0 Å) were obtained for 8 out of 11 proteins using comparative modeling.
• For proteins without sequence similarity, fragment assembly yielded models with RMSD ~3.0 Å for 2 cases.
• The scoring function consistently identified the native structure as the best-scored structure among all predictions.
• No significant RMSD improvement was achieved through molecular dynamics refinement of predicted structures.

Conclusions:

• The developed free energy function demonstrates high reliability for protein structure prediction.
• Current sampling methods, particularly for proteins lacking sequence homology, represent a significant bottleneck.
• Further advancements in sampling strategies are essential for enhancing the accuracy of protein structure prediction.