Computational Approaches for Drug Repositioning

Abstract

Drug development is a costly and time consuming activity. The traditional process relies on extensive experimental efforts to map out the relevant part of the chemical space. Data about molecules, diseases, genes and other entities are present on many isolated databases, be that internal or external and in heterogeneous formats. They either require costly and inflexible data integration, or time-consuming workflows. Computational approaches, and more recently artificial intelligence based techniques, have emerged as a promising alternative for reducing the development cycle through drug repositioning. Knowledge bases are used to predict new links between old drugs and new targets. We present below the overall approach adopted for my PhD thesis, for a more holistic knowledge graph-based drug repositioning that aims to discover hidden or missing links between existing drugs and targets for which no known treatment is available. Currently, eight data and knowledge resources have already been integrated into the designed knowledge graph.