SynDRep: a synergistic partner prediction tool based on knowledge graph for drug repurposing.

Area of Science:

• Pharmacology
• Bioinformatics
• Computational Biology

Background:

• Drug repurposing offers a cost-effective and low-risk strategy for developing new therapeutics.
• Existing drug repurposing methods often overlook the potential of drug combination data, focusing instead on single drug-disease-target relationships.

Purpose of the Study:

• To introduce SynDRep, a novel computational tool for drug repurposing.
• To leverage knowledge graphs enriched with drug combination effects for predicting synergistic drug partners.
• To repurpose identified partner drugs as single agents for target diseases.

Main Methods:

• Enrichment of knowledge graphs (KGs) with drug combination effects.
• Application of graph embedding techniques (e.g., HolE) for relationship prediction.
• Utilizing machine learning (ML) models (e.g., random forest) for drug repurposing predictions.

Main Results:

• The HolE embedding model achieved 84.58% accuracy in predicting relationships.
• Random forest model demonstrated strong performance with an ROC-AUC of 0.796.
• Identified potential drug candidates, such as miconazole and albendazole for Alzheimer's disease, with some literature validation.

Conclusions:

• SynDRep provides a novel framework for drug repurposing by incorporating drug combination insights.
• The tool shows promise in identifying novel therapeutic applications for existing drugs.
• Further enhancements with specialized KGs and additional data can improve SynDRep's efficacy for cost-effective drug discovery.