MEGADOCK-Web

MEGADOCK-Web provides high-throughput structure-based predictions of protein–protein interactions to support exploration of cellular functions and inhibitor design.

Key Features:

• Extensive database: Contains 7,528 protein chains and more than 28 million predicted PPIs generated from all pairwise combinations of the included proteins.
• Structure-based docking predictions: High-throughput structure-based docking was used to produce precalculated PPI predictions for the protein set.
• Precalculated PPI predictions: A repository of precomputed docking results for all pairwise protein combinations is maintained.
• Annotations: Each protein structure is annotated with PDB ID, chain ID, UniProt AC, and related KEGG pathway IDs.
• Known PPI annotation: Predicted protein pairs are flagged with experimentally known PPI pairs when such annotations exist.
• Biochemical pathway visualization: Predicted interactions can be mapped onto biochemical pathways (KEGG) to show potential pathway contexts.
• 3D molecular visualization: Predicted complex structures are provided as 3D models for structural assessment of interactions.

Scientific Applications:

• Discovery of novel PPIs: Identify potential protein–protein interactions beyond experimentally characterized pairs.
• Functional inference: Explore cellular functions by mapping predicted interactions to KEGG biochemical pathways.
• Inhibitor design support: Use predicted complex structures to inform design of molecules that disrupt PPIs.
• Validation and prioritization: Compare predicted PPIs with annotated known interactions to prioritize candidates for experimental follow-up.
• Structural feasibility assessment: Evaluate the plausibility of predicted interactions using 3D predicted complex models.

Methodology:

High-throughput structure-based docking was performed on all pairwise combinations of 7,528 protein chains to generate >28 million predicted PPIs; results are annotated with PDB ID, chain ID, UniProt AC, KEGG pathway IDs and marked where experimentally known PPI pairs exist, and predicted complexes are provided as 3D structures.