Generating new drug repurposing hypotheses using disease-specific hypergraphs.

Area of Science:

• Computational biology
• Network medicine
• Pharmacology

Background:

• Drug development is lengthy and expensive; drug repurposing offers a cost-effective alternative.
• Current network-based computational methods for drug repurposing lack disease specificity, potentially limiting their effectiveness.
• Existing interactome models aggregate disease and interaction data, diluting information crucial for identifying relevant drug targets.

Purpose of the Study:

• To develop a disease-specific computational approach for enhanced drug repurposing.
• To address the limitations of aggregate network models by incorporating disease-specific biological pathways.
• To identify novel drug repurposing candidates for Alzheimer's disease (AD) with supporting biological rationale.

Main Methods:

• Constructed disease-specific hypergraphs where hyperedges represent biological pathways.
• Utilized a modified node2vec algorithm to generate pathway embeddings.
• Evaluated the hypergraph approach for AD drug repurposing, comparing results against a state-of-the-art knowledge graph (multiscale interactome).

Main Results:

• Identified 7 promising drug repurposing candidates for Alzheimer's disease.
• These candidates were ranked as unlikely by the multiscale interactome but supported by existing literature.
• The method provided explanations for drug repositioning suggestions, highlighting plausible biological pathways.

Conclusions:

• Disease-specific hypergraphs represent a significant advancement in computational drug repurposing.
• The proposed method successfully identifies novel therapeutic opportunities for complex diseases like AD.
• Future work includes scaling the method for broader disease coverage and personalized recommendations based on patient comorbidities.