The linear interaction energy method for the prediction of protein stability changes upon mutation

Area of Science:

• Computational Biology
• Biophysics
• Protein Engineering

Background:

• Understanding the link between protein sequence and energetics is crucial for computational protein design.
• Accurate and efficient computational tools are needed to analyze large sequence datasets for insights into protein evolution, disease, and drug resistance.

Purpose of the Study:

• To develop and validate a computational approach for predicting free-energy changes in proteins due to single mutations.
• To assess the accuracy and applicability of the Linear Interaction Energy (LIE) approximation for this purpose.

Main Methods:

• Utilized the Linear Interaction Energy (LIE) approximation.
• Applied the method to predict free-energy changes for 822 mutations across 10 different proteins.
• Analyzed the correlation between calculated and experimental ΔΔG values.

Main Results:

• Achieved an average unsigned error of 0.82 kcal/mol.
• Obtained a correlation coefficient of 0.72 between predicted and experimental ΔΔG values.
• Successfully identified destabilizing hot spot mutations but showed limitations in distinguishing between stabilizing and destabilizing mutations.

Conclusions:

• The LIE-based approach provides a computationally inexpensive and accurate method for predicting mutation-induced free-energy changes.
• The model shows promise for further investigations into protein stability, fitness, correlated mutations, and drug resistance.
• Further refinement may be needed to improve the distinction between stabilizing and destabilizing mutations.