Protein structure prediction assisted with sparse NMR data in CASP13.

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Protein structure prediction is crucial for understanding biological function.
• Sparse Nuclear Magnetic Resonance (NMR) data, including NOESY and residual dipolar couplings, is often limited for larger proteins.
• The CASP13 (Critical Assessment of protein Structure Prediction) competition provided a benchmark for evaluating prediction methods.

Purpose of the Study:

• To assess the impact of sparse NMR data on the accuracy of protein structure prediction.
• To compare NMR-assisted prediction methods against traditional, non-assisted approaches.
• To explore the utility of NMR data for validating and refining predicted protein structures.

Main Methods:

• Simulated sparse NMR data (NOESY, ¹⁵N-¹H residual dipolar couplings) for 11 CASP13 targets (80-326 residues).
• Utilized real NMR data, including a dataset with only backbone assignments, for de novo designed proteins.
• Compared models generated by NMR-assisted prediction groups with those from regular, non-assisted prediction groups.

Main Results:

• NMR-assisted prediction groups generated more accurate models for several targets compared to baseline methods.
• Incorporation of sparse, often noisy, NMR data generally resulted in higher accuracy models.
• For six out of 13 targets, the best NMR-assisted model outperformed the best regular prediction; however, regular methods were superior for the remaining seven targets.

Conclusions:

• Sparse NMR data can significantly improve protein structure prediction accuracy when integrated with advanced computational techniques.
• A hybrid approach, using prediction methods to generate initial models followed by NMR data for refinement, offers a novel strategy for protein structure determination.
• Further development is needed to fully leverage NMR data for enhancing protein structure prediction, especially when compared to state-of-the-art non-assisted methods.