CoRNeA

CoRNeA predicts protein-protein interaction interfaces in eukaryotic complexes from amino acid sequence information using co-evolution analysis, Random Forest classification, and network analysis.

Key Features:

• Hybrid Framework: Integrates co-evolution analysis, Random Forest machine learning, and network analysis to improve interface prediction accuracy.
• Co-Evolution Analysis: Examines co-evolutionary patterns among amino acid residues to identify potential interface residues.
• Random Forest Machine Learning: Employs a Random Forest classifier trained on features derived from co-evolutionary signals, physicochemical properties, and contact potentials to distinguish interface from non-interface residues.
• Network Analysis: Contextualizes predicted interfaces within interaction networks to provide insights into functional implications.
• Intra-Contact Information Integration: Incorporates intramolecular contact information to reduce false positives by accounting for residues involved in intra-protein folding.
• Sequence-based, Eukaryote-specific Modeling: Operates on amino acid sequence data and accounts for eukaryote-specific composition and evolutionary rates.

Scientific Applications:

• Functional annotation of eukaryotic protein complexes: Predicts interface residues to support interpretation of complex function.
• Structural biology: Maps likely interface regions to inform structure interpretation and modeling efforts.
• Drug discovery: Identifies interface residues that can inform strategies targeting protein–protein interactions.
• Systems biology: Integrates predicted interfaces into interaction networks to study cellular machinery and pathways.

Methodology:

Performs pairwise analysis of amino acid sequences, derives co-evolutionary features, physicochemical properties, contact potentials and intra-contact information, trains a Random Forest classifier on these features, and applies network analysis to contextualize predicted interfaces.