Correlated mutations select misfolded from properly folded proteins.

Area of Science:

• Computational Biology
• Structural Bioinformatics

Background:

• Direct coupling analysis (DCA) has advanced residue-residue contact prediction from multiple sequence alignments.
• Current contact prediction accuracy remains a limitation for *ab initio* protein structure prediction.
• DCA offers potential support for various protein structure studies.

Purpose of the Study:

• To evaluate the capability of DCA in distinguishing between correctly folded and misfolded protein structures.
• To assess DCA's performance using existing Protein Data Bank (PDB) files and decoy models.

Main Methods:

• Comparison of DCA predictions for correctly folded protein structures versus misfolded decoy models.
• Analysis of DCA performance on X-ray crystallography and Nuclear Magnetic Resonance (NMR) structures.

Main Results:

• DCA systematically predicts more contacts for properly folded structures compared to misfolded ones.
• DCA successfully selected the correct structure from folded and misfolded variants.
• The method demonstrated higher efficacy with X-ray structures than with NMR structures.

Conclusions:

• DCA is a valuable tool for validating protein structure models by identifying misfolded variants.
• The findings support the utility of DCA in refining and assessing protein structures.
• Further application of DCA can enhance the accuracy of protein structure determination.