Can disease-specific hypergraphs help generate drug repurposing hypotheses?

Original title: Generating Drug Repurposing Hypotheses through the Combination of Disease-Specific Hypergraphs

Authors: Ayush Jain, Marie Laure-Charpignon, Irene Y. Chen, Anthony Philippakis, Ahmed Alaa

In this article, the author discusses the lengthy and expensive process of developing new drugs and proposes an alternative approach called drug repurposing. They explain that computational techniques based on biomedical knowledge graphs have led to new hypotheses for repurposing drugs. The author introduces a new technique that uses hypergraph representations to analyze biological pathways and derive embeddings that link specific drugs to diseases of interest. They further expand this method to include multiple diseases. By calculating cosine similarity values, the author ranks 1,522 drugs based on their repurposing potential. The article focuses on Alzheimer’s disease (AD) and its risk factors, hypertension (HTN) and type 2 diabetes (T2D), and compares the rankings of drugs in different hypergraph settings. The author identifies two promising drugs, dapagliflozin and debrisoquine, whose repurposing potential significantly increased when considering combinations of diseases. The article concludes by emphasizing that this approach is a tool for generating hypotheses and should be validated through laboratory experiments and analysis of biomedical literature.

Original article: https://arxiv.org/abs/2311.09596