PDID

PDID provides structure-based predictions of drug–protein interactions to identify putative binding sites and off-targets in the human proteome.

Key Features:

• Extensive Coverage: PDID encompasses 9,652 protein structures from 3,746 proteins, covering a broad portion of the human proteome.
• High-Volume Predictions: The database contains approximately 16,800 putative interactions derived from nearly 1.1 million structure-based predictions for several popular drugs.
• Advanced Prediction Methods: PDID uses three computational methods—ILbind, SMAP, and eFindSite—to generate all-atom structure-based interaction predictions.
• Propensity Scores: Each predicted interaction is accompanied by a propensity score that quantifies the likelihood of the interaction occurring.
• Binding Coordinates: PDID provides detailed coordinates for putative drug binding locations within protein structures to support structural analyses.
• Integration with Curated Data: Predicted interactions are complemented by experimentally curated interactions from DrugBank, BindingDB, and the Protein Data Bank (PDB).

Scientific Applications:

• Polypharmacology: Identification of off-target effects to study multi-target drug actions and potential adverse side effects.
• Drug repurposing: Revealing potential new targets for existing drugs to support repurposing efforts.
• Structural and functional analysis: Mapping putative binding coordinates within protein structures to aid structural interpretation and functional hypothesis generation.

Methodology:

Predictions are generated using ILbind, SMAP, and eFindSite with results evaluated by propensity scoring and mapped to binding coordinates within protein structures.