MuToN Quantifies Binding Affinity Changes upon Protein Mutations by Geometric Deep Learning.

Area of Science:

• Computational biology
• Structural biology
• Biophysics

Background:

• Understanding protein-protein binding affinity changes due to mutations is vital for cellular processes.
• Accurate computational prediction of these affinity changes remains a significant challenge due to complex biological mechanisms.

Purpose of the Study:

• To introduce MuToN, a novel geometric deep learning framework for quantifying protein binding affinity changes upon residue mutations.
• To develop a mechanism-aware method that captures interface alterations and allosteric effects.

Main Methods:

• Utilized geometric attention networks within a deep learning framework.
• Designed the method to analyze changes in protein binding interfaces of mutated complexes.
• Incorporated assessment of allosteric effects of amino acids.

Main Results:

• MuToN demonstrated superior performance compared to existing computational methods.
• The framework accurately predicted binding affinity changes for SARS-CoV-2 variants interacting with the ACE2 complex.
• Experimental results validated MuToN's effectiveness and flexibility.

Conclusions:

• MuToN offers a powerful and accurate approach for predicting mutation-induced binding affinity changes.
• The method's mechanism-aware design enhances its predictive capabilities.
• MuToN's application to viral variants highlights its practical relevance in understanding infectious diseases.