iStable

iStable predicts changes in protein stability resulting from single amino acid mutations by integrating sequence information and outputs from multiple element predictors.

Key Features:

• Integrated Prediction Approach: Combines outputs from multiple element predictors and integrates them rather than relying on majority voting.
• Machine Learning Integration: Employs support vector machines (SVM) as the integrator after evaluation of several machine learning methods.
• Sequence Information Analysis: Incorporates sequence context using an optimal 11-residue window for prediction.
• Dual Input Types: Accepts both structural and sequential input types for analysis.
• Grid Computing Architecture: Utilizes a grid computing architecture to enhance computational integration of predictor results.
• Comprehensive Validation: Uses training and cross-validation and reports improved performance across different secondary structures, relative solvent accessibility conditions, protein superfamilies, and experimental scenarios.

Scientific Applications:

• Protein engineering and design: Predicts stability changes to guide selection and optimization of mutation candidates for novel protein designs.
• Mutation effect assessment: Assesses the impact of single amino acid substitutions on protein stability across diverse structural contexts.
• Comparative analysis across protein families: Enables evaluation of stability prediction performance across protein superfamilies and experimental conditions.

Methodology:

Integrates sequence information with outputs from multiple element predictors using a grid computing architecture and an SVM-based integrator, employing an 11-residue window for sequence context and validated via training and cross-validation.