DeepComplex

DeepComplex predicts quaternary structures of dimeric protein complexes by using deep learning to infer inter-chain contacts and distance-based modeling to assemble dimer quaternary structures.

Key Features:

• Inter-chain contact prediction: Predicts inter-chain contacts for homodimers and heterodimers.
• Deep learning algorithm: Employs a deep learning model trained to identify inter-chain interactions from tertiary structures.
• Distance-based modeling: Constructs quaternary dimer structures using distance-based modeling informed by predicted contacts.
• Input requirements: Accepts tertiary structure data for one chain (homodimers) or two chains (heterodimers).
• Predicted contact maps: Produces predicted inter-chain residue-residue contact maps.
• Multiple sequence alignments: Generates multiple sequence alignments associated with predicted contacts.
• Modeled quaternary structures: Outputs modeled quaternary structures of the dimer.

Scientific Applications:

• Protein function analysis: Enables investigation of protein interaction interfaces and quaternary organization to infer functional mechanisms.
• Protein engineering: Facilitates design and modification of protein interfaces by providing structural models of dimeric assemblies.
• Drug design: Supports identification of potential therapeutic targets and interface hotspots by modeling protein-protein interactions at the quaternary level.

Methodology:

Deep learning-based prediction of inter-chain contacts from tertiary structures followed by distance-based modeling to assemble the quaternary dimer structure.