SAAMBE-3D

SAAMBE-3D predicts the effects of single amino acid mutations on protein-protein binding free energy and classifies mutations as disruptive or non-disruptive to support analysis of disease-associated interaction changes.

Key Features:

• Machine Learning-Based Approach: Employs a machine learning methodology to predict mutation-induced changes in protein-protein binding free energy.
• Dual Prediction Capabilities: Calculates binding free energy changes (ΔΔG) caused by single amino acid mutations and classifies mutations as disruptive or non-disruptive to protein interactions.
• High Accuracy: Benchmarking against the SKEMPI v2.0 database yields Pearson correlation coefficients of 0.78–0.82 and Area Under Curve (AUC) values of 1.0 for homo-dimer datasets and 0.96 for hetero-dimer datasets.
• Rapid Processing: Completes predictions in less than a fraction of a second, enabling large-scale analyses.
• Optimization and Validation: Methodology optimized and validated using five-fold cross-validation on datasets from Cornell University.

Scientific Applications:

• Disease mechanism analysis: Interprets how amino acid substitutions alter protein-protein interactions implicated in disease.
• Genome-wide and large-scale genetic studies: Supports GWAS and other large-scale analyses for identifying disease-associated interaction-disrupting mutations.

Methodology:

Machine learning-based prediction of ΔΔG and binary classification of disruptive versus non-disruptive mutations; benchmarking against the SKEMPI v2.0 database; five-fold cross-validation on datasets from Cornell University with reported Pearson correlation coefficients and AUC metrics.