An update on knowledge graphs and their current and potential applications in drug discovery

Knowledge graphs are becoming prominent tools in computational drug discovery. They effectively integrate heterogeneous biomedical data and generate new hypotheses and knowledge.

This article is based on a literature review using Google Scholar and PubMed to retrieve articles on existing knowledge graphs relevant to the drug discovery field. The authors compare the types of entities, relationships, and data sources they encompass. Additionally, the authors provide examples of their use in the drug discovery field and discuss potential strategies for advancing this research area.

Knowledge graphs are crucial in drug discovery, but their construction leads to challenges in data integration and consistency. Future research should prioritize the standardization of data sources and data modeling. More efforts are needed for the integration in knowledge graphs of diverse data types, such as chemical structures and epigenetic data, to enhance their effectiveness. Additionally, advancements in large language models should be pursued to aid the development of knowledge graphs, provide intuitive querying capabilities for non-expert users, and explain knowledge graphs -derived predictions, thereby making these tools more accessible and their insights more interpretable for a wider audience.