Predicting mutations deleterious to function in beta-lactamase TEM1 using MM-GBSA.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Genetics and Genomics

Background:

• Missense mutations can significantly alter protein function, leading to various diseases.
• Differentiating between functionally disruptive and neutral missense variants is vital for drug discovery and diagnostics.
• Statistical inference of mutation impact is challenging due to the rarity of specific variants.

Purpose of the Study:

• To benchmark a physics-based approach, MM-GBSA, for predicting functionally disruptive missense mutations.
• To compare MM-GBSA performance against alternative prediction strategies.
• To assess the generalizability of the MM-GBSA approach across diverse mutation datasets.

Main Methods:

• Utilized a large dataset of 990 residue mutations in beta-lactamase TEM1.
• Applied the MM-GBSA (Molecular Mechanics with Generalized Born Surface Area) method to predict changes in protein stability.
• Compared MM-GBSA predictions with other computational strategies for mutation classification.

Main Results:

• MM-GBSA achieved an Area Under the Curve (AUC) of 0.75 on the beta-lactamase TEM1 dataset.
• The MM-GBSA method outperformed all other tested predictors.
• Performance remained robust across different dataset divisions, indicating generalizability, with surface mutations posing the greatest classification challenge.

Conclusions:

• The physics-based MM-GBSA approach is a powerful and generalizable tool for classifying functionally disruptive missense mutations.
• MM-GBSA offers improved accuracy compared to existing methods for mutation impact prediction.
• Further research is needed to address the challenges in classifying the impact of surface protein mutations.