mCSM-membrane

mCSM-membrane predicts the effects of single-point mutations on transmembrane proteins by assessing mutation-induced changes in protein stability and pathogenicity.

Key Features:

• Graph-based signature approach: Represents protein geometry and physicochemical properties with graph-based signatures for mutation modelling.
• Supervised learning: Uses supervised learning models to predict changes in stability and to classify mutation pathogenicity.
• Performance metrics: Stability predictor achieves correlations up to 0.72 in cross-validation and 0.67 in blind tests; pathogenicity predictor achieves Matthew's Correlation Coefficient (MCC) up to 0.77 in cross-validation and 0.73 in blind tests.

Scientific Applications:

• Structural and functional impact analysis: Predicts how single-point mutations affect the structure and function of transmembrane proteins.
• Disease variant identification: Identifies potentially disease-associated variants in membrane proteins by assessing pathogenicity.
• Experimental planning: Prioritizes mutations to guide empirical validation and experimental design.

Methodology:

Represents protein geometry and physicochemical properties using graph-based signatures and applies supervised learning models, with performance evaluated by cross-validation and blind tests.