CASP11 refinement experiments with ROSETTA

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• Protein structure refinement is crucial for accurate biological function prediction.
• Existing methods face challenges in handling diverse starting model qualities.
• The Critical Assessment of protein Structure Prediction (CASP) experiment provides a benchmark for evaluating new methods.

Purpose of the Study:

• To develop and test novel Rosetta-based computational approaches for protein structure refinement.
• To evaluate the performance of different refinement strategies based on initial model quality.
• To identify strengths and weaknesses of automated refinement protocols in a large-scale experiment.

Main Methods:

• Development of automated refinement protocols integrating sampling, parallel computation, and multiobjective optimization.
• Application of two distinct strategies in CASP11: aggressive large-scale rebuilding for poor models and local rebuilding with core refinement for better models.
• Adaptive refinement based on the predicted accuracy of starting structures.

Main Results:

• The high-resolution refinement strategy (local rebuilding + core refinement) consistently improved initial protein models.
• The low-resolution strategy (large-scale rebuilding) showed variable success, with frequent worsening of models due to selection issues.
• Performance varied significantly with the quality and predicted accuracy of the starting structures.

Conclusions:

• The high-resolution refinement protocol shows promise as a complementary method for protein structure refinement.
• The low-resolution refinement strategy requires substantial further development to overcome model selection challenges.
• Automated refinement protocols can be tailored based on initial model quality for improved performance.