A Network Medicine Approach to Drug Repurposing for Chronic Pancreatitis

Despite decades of clinical investigations, there is currently no effective treatment for patients diagnosed with Chronic Pancreatitis (CP). Computational drug repurposing holds promise to rapidly identify therapeutics which may prove efficacious against the disease. Using a literature-derived knowledge graph, we train multiple machine learning models using embeddings based on i) the network topology of regulation bipartite networks, ii) protein primary structures and iii) molecule substructures. Using these models, we predict approved drugs that down-regulate the disease, and assess their proposed respective drug targets and mechanism of actions. We analyse the highest predicted drugs and find a diverse range of regulatory mechanisms including inhibition of fibrosis, inflammation, immmune response, oxidative stress and calcium homeostasis. Notably, we identify resiniferatoxin, a potent analogue of capsaicin, as a promising repurposable candidate due to its antiinflammatory properties, nociceptive pain suppression, and regulation of calcium homeostatis (through potentiation of mutant cystic fibrosis transmembrane conductance regulator (CFTR)). Resiniferatoxin may also regulate intracellular acinar Ca2+ via agonism of transient receptor potential vanilloid subfamily member 6 (TRPV6). We believe the potential of this repurposable drug warrants further in silico and in vitro testing, particularly the affect of the TRPV6 agonism on disease pathogenesis.