DDGun

DDGun predicts changes in protein stability expressed as the difference in Gibbs free energy of unfolding (∆∆G) caused by single or multiple amino acid mutations using evolutionary information.

Key Features:

• Untrained methodology: Operates without training on experimental datasets and relies on basic features derived from evolutionary information, enabling its use as a benchmark for evaluating individual predictive features and supervised methods.
• Anti-symmetric prediction property: Ensures that the predicted ∆∆G for a mutation A → B equals the negative of the predicted ∆∆G for the reverse mutation B → A.
• Performance: Achieves Pearson correlation coefficients of ≈0.5 for single-site variations and ≈0.4 for multiple-site variations from sequence data (DDGun), with slightly higher correlations when structure is included (DDGun3D).
• Versatility: Predicts stability changes for both single-site and multiple-site amino acid substitutions.

Scientific Applications:

• Protein engineering: Guides design by predicting the thermodynamic effects of amino acid substitutions on protein stability.
• Disease mutation analysis: Assesses the stability impact of genetic variants implicated in disease phenotypes.
• Drug design: Informs strategies that modulate protein stability relevant to therapeutic development.
• Structure–function studies: Provides quantitative ∆∆G estimates to relate sequence variation to structural and functional changes.

Methodology:

Uses basic features derived from evolutionary information to predict ∆∆G and is implemented in two versions—sequence-only (DDGun) and sequence-plus-structure (DDGun3D); performance has been validated through blind tests.