CoABind

CoABind predicts coenzyme A (CoA) and CoA-derivative binding residues in proteins to identify interaction sites relevant to metabolic regulation and functional annotation.

Key Features:

• SVMpred: Uses complementary sequence profiles, predicted secondary structures, and solvent accessibility with support vector machines for residue-level classification.
• TemPred: Identifies homologous templates with HHsearch and transfers CoA-binding annotations from known CoA-binding proteins.
• Consensus integration: Integrates SVMpred and TemPred predictions to combine ab initio and template-based evidence.
• Input features: Incorporates sequence profiles, predicted secondary structure, and solvent accessibility as predictive features.
• Performance: Achieves a Matthews correlation coefficient (MCC) of 0.489 on an independent test set of 73 proteins.

Scientific Applications:

• Mapping CoA-binding residues: Enables identification of CoA and CoA-derivative interaction sites to study CoA–protein interactions.
• Functional annotation: Supports annotation of protein binding sites for structural, mechanistic, and metabolic pathway studies.
• Benchmarking: Provides specialized CoA-binding predictions for comparison with general-purpose ligand-binding predictors such as COACH.

Methodology:

SVMpred applies support vector machines using sequence profiles, predicted secondary structure, and solvent accessibility; TemPred transfers binding annotations from homologous templates identified by HHsearch; a consensus combines SVMpred and TemPred predictions; evaluation reports MCC = 0.489 on an independent test set of 73 proteins.