redrugs

is a Knowledge Graph.

ReDrugs is a probabilistic knowledge graph for drug repositioning that integrates drug-target, protein-protein, and disease-gene interactions from multiple databases. The system uses evidence-weighted nanopublications to assign confidence scores to interactions based on experimental methods and manual curation. ReDrugs was designed to identify novel drug candidates for diseases, particularly melanoma, by filtering and analyzing systems biology networks with probabilistic methods. The platform included both a web interface and API for exploring molecular interaction networks. The web interface appears to be no longer accessible.

Compare

Infores ID

Unknown

FAIRsharing ID

Unknown

Product Summary

Publications

Finding melanoma drugs through a probabilistic knowledge graph

Products

From this Resource

ID	Name	URL	Category	Format	Description
redrugs.web	ReDrugs Web Interface	redrugs.tw.rpi.edu	GraphicalInterface	❔	Interactive web interface for explori...
redrugs.api	ReDrugs API	api	ProgrammingInterface	❔	SADI web services API for querying th...

Relevant Taxa

Human (NCBITaxon:9606)

Details

ReDrugs

Overview

ReDrugs (Drug Repositioning through Semantic Integration) is a probabilistic knowledge graph platform designed to identify drug repositioning candidates by integrating and analyzing molecular interactions from multiple biological databases. The system combines drug-target interactions from DrugBank, protein-protein interactions from iRefIndex, gene ontology annotations from UniProt GOA, and disease-gene associations from OMIM and the COSMIC gene census.

The platform’s key innovation is its use of nanopublications with evidence-based probabilistic scoring. Each interaction assertion is assigned a confidence probability based on either manual curation (p=0.999) or the quality of the experimental method used (ranging from p=0.8 to p=0.99). These probabilities are combined using composite Z-scores to provide overall confidence scores for complex interaction pathways.

ReDrugs was successfully applied to melanoma drug discovery, identifying 25 high-quality drug candidates with a joint probability ≥0.93 and ≤3 interaction steps from the disease. The system validated well, with nearly all predicted drugs having evidence from clinical trials or experimental studies.

Note: The ReDrugs web interface and API appear to be no longer accessible as of 2025.

Key Features

Evidence-Weighted Knowledge Graph: 6,180 drugs, 3,820 diseases, 69,279 proteins, and 899,198 interactions
Probabilistic Filtering: Joint probability calculations for multi-step interaction paths
Semantic Web Standards: Uses RDF, OWL, SPARQL, PROV-O, and SIO ontologies
Interactive Visualization: Cytoscape.js-based network exploration with filtering controls
Nanopublication Architecture: Each interaction captured with assertion, provenance, and publication metadata

Data Sources

DrugBank: Drug-target interactions (manual curation)
iRefIndex: Protein-protein interactions and complexes
UniProt GOA: Gene ontology annotations
OMIM: Disease-gene associations
COSMIC Gene Census: Cancer-gene associations

Technical Architecture

Backend: Blazegraph RDF database with SPARQL endpoint
Web Framework: Python TurboGears with SADI web services
Frontend: AngularJS with Cytoscape.js visualization
Data Format: RDF nanopublications with PSI-MI and SIO ontologies

Automated Evaluation

View the automated evaluation: redrugs automated evaluation