ProTSPoM

ProTSPoM predicts alterations in Gibbs free energy caused by missense mutations to evaluate effects on protein stability by integrating residue properties, fold-level attributes, environmental compatibility, and evolutionary information using Random Forest and Gradient Boosted Regressors.

Key Features:

• Machine learning models: Employs Random Forest Regressors and Gradient Boosted Regressors for regression of Gibbs free energy changes due to missense mutations.
• Input feature types: Integrates residue properties, fold-level attributes, environmental compatibility metrics, and evolutionary information as predictive features.
• Prediction target: Predicts alterations in Gibbs free energy (stability changes) resulting from single amino-acid substitutions.
• Benchmark performance: Reports Pearson correlation coefficients of 0.82 for S350 and 0.88 for p53 with root-mean-squared-errors of 0.92 kcal/mol for S350 and 1.06 kcal/mol for p53.

Scientific Applications:

• p53 protein analysis: Identifies missense mutations that affect the structural integrity and DNA binding affinity of p53.

Methodology:

Uses Random Forest and Gradient Boosted Regressors to integrate residue properties, fold-level attributes, environmental compatibility, and evolutionary information to predict Gibbs free energy changes from missense mutations.