Refining homology models by combining replica-exchange molecular dynamics and statistical potentials.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Modeling

Background:

• Homology modeling is crucial for predicting protein structures.
• Refining these models is essential for accurate functional and structural insights.
• Current refinement methods face limitations in sampling and scoring.

Purpose of the Study:

• To develop and validate a protocol for global refinement of protein homology models.
• To combine replica-exchange molecular dynamics (REMD) for conformational sampling with statistical potentials for model selection.
• To assess the effectiveness of the protocol in improving the accuracy of homology models.

Main Methods:

• Utilized temperature-based replica-exchange molecular dynamics (REMD) for extensive conformational sampling.
• Employed statistical potentials for ranking and selecting refined protein models.
• Tested the protocol on 21 homology models, including those of small proteins with known crystal structures and targets from the CASPR exercise.

Main Results:

• REMD successfully sampled near-native conformational states from high-quality homology models.
• Achieved improvements in secondary structure element root-mean-square deviation (SSE-RMSD) ranging from 0.5-1.0 Å for 15 out of 21 cases.
• Statistical potentials identified improved structures within the top-ranked models, with an average SSE-RMSD improvement of 0.42 Å.

Conclusions:

• The proposed protocol effectively refines high-quality homology models for small proteins.
• Scoring functions represent a significant limitation in current structure refinement methodologies.
• Further improvements in scoring functions are necessary to enhance the accuracy of protein structure prediction and refinement.