Exploring the utility of link prediction algorithms for drug repurposing
Tiffany J. Callahan, William A. Baumgartner, Adrianne L. Stefanski, Larry E. Hunter, Olivier Bodenreider · 2019
It takes roughly 10-12 years and costs $2.87 billion to bring a new drug to market. Further, of all funded pre-clinical trials, only 10% will actually advance to human testing. A limited repertoire of valid biomarkers, few validated clinical phenotypes, and an incomplete understanding of biological mechanisms contribute to the delay of novel pharmaceutical development. Network representations facilitate the understanding of complex biological mechanisms, and have been used extensively in biomedical research. Link prediction, a network inference method used to predict new connections between nodes in a network, has correctly predicted important relationships, including protein-protein and regulatory gene interactions. OBJECTIVE: To explore the utility of link prediction algorithms for identifying novel drug-target interactions.