Local interactions in protein folding determined through an inverse folding model

Area of Science:

• Computational Biology
• Protein Folding
• Bioinformatics

Background:

• Protein folding stability is influenced by hydrophobic effects and local interactions governing secondary structure.
• Accurate prediction of protein structures requires precise parameterization of these interactions.
• Traditional methods using amino acid propensities can be influenced by secondary structure burial preferences.

Purpose of the Study:

• To develop and optimize an inverse folding model incorporating both hydrophobic and local interactions.
• To determine optimal local interaction parameters by fitting the model to empirical data from the Protein Data Bank.
• To evaluate the model's performance in recognizing native and similar protein structures.

Main Methods:

• An inverse folding model was developed, combining hydrophobic effects and local interactions for secondary structure.
• Optimal local interaction parameters were determined by fitting the model to site-specific amino acid distributions from the Protein Data Bank.
• The model's accuracy was assessed using likelihood, Bayesian Information Criterion, and threading experiments, including challenging homologous structure recognition.

Main Results:

• The optimized inverse folding model achieved a mean correlation coefficient of 0.96 between observed and predicted amino acid distributions.
• The model correctly identified the native structure in 100% of challenging threading tests.
• Local interactions derived from inverse folding significantly improved the discrimination of native and similar structures compared to propensity-based or contact-only methods.

Conclusions:

• Optimized local interaction parameters derived from inverse folding substantially enhance protein structure prediction and recognition.
• The developed model provides a more accurate representation of forces governing protein folding stability.
• Further research should focus on challenging tests, such as homologous structure recognition, to rigorously evaluate protein folding stability scores.