Knowledge-based entropies improve the identification of native protein structures.

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Accurate protein structure evaluation necessitates reliable free energy calculations, encompassing both potential energies and entropies.
• While knowledge-based potentials have shown success, computing protein entropies remains a significant challenge.

Purpose of the Study:

• To develop and evaluate knowledge-based conformational entropies for proteins.
• To assess the impact of these entropies on protein structure prediction and model assessment.

Main Methods:

• Analyzed contact changes between amino acids in 167 diverse proteins with two alternative structures.
• Derived knowledge-based entropies using the inverse Boltzmann relationship, analogous to knowledge-based potentials.
• Correlated contact change patterns with amino acid properties like solvent exposure and polarity.

Main Results:

• Charged and polar interactions were found to break more frequently than hydrophobic pairs.
• Observed patterns strongly correlated with amino acid solvent exposure, polarity, and size.
• Incorporating these knowledge-based entropies nearly doubled the performance in native protein structure selection.

Conclusions:

• The developed knowledge-based entropies offer a novel and effective approach to protein structure evaluation.
• Energy-entropy compensation observed at both global and pairwise levels.
• These entropies have direct applications in 3D structure prediction, protein model assessment, and protein engineering.