ComplexContact

ComplexContact predicts interfacial residue-residue contacts between two protein sequences to identify interacting residues in heterodimers and support residue-level analysis of protein interactions.

Key Features:

• Interfacial residue-residue contact prediction: Predicts how residues from different proteins interact at their interface to identify inter-protein contacts.
• Sequence-based paired MSA construction: Accepts a pair of protein sequences and identifies sequence homologs to construct paired multiple sequence alignments (MSAs).
• Co-evolution analysis: Analyzes co-evolutionary patterns within paired MSAs to infer potential residue-residue contacts.
• Deep learning methodology: Applies an ultra-deep learning model originally developed for intra-protein contact prediction and recognized as a CASP12 winner to inter-protein predictions.
• Visualization: Outputs predicted contacts as images representing residue-residue contact maps.
• Performance: Outperforms pure co-evolution methods in inter-protein contact prediction across multiple species.

Scientific Applications:

• Heterodimer contact prediction: Predicts interfacial contacts between two protein chains in heterodimeric complexes.
• Protein complex formation analysis: Informs understanding of residue-level interactions involved in protein complex assembly.
• Cross-species contact analysis: Enables comparative inter-protein contact prediction across different species.

Methodology:

Constructs paired multiple sequence alignments from sequence homologs, performs co-evolution analysis on the MSAs, and applies an ultra-deep learning model (CASP12 winner) for inter-protein contact prediction.