Understanding drug mechanisms is essential for comprehending how drugs interact with biological targets, enabling the prediction of their efficacy and potential side effects. However, existing databases are limited in the number of substances with detailed drug mechanisms, making comprehensive drug analysis challenging. Additionally, drug mechanisms are often provided in textual form, presenting issues of accuracy and efficiency in extraction and integration processes. Therefore, we propose DrugMecha, a knowledge graph database that extracts and integrates drug mechanism information from the PubChem Bioassay database. DrugMecha employs advanced text mining tools to extract information on entities and their relationships, including Molecular Functions, Genes, Biological Processes, Cell Lines, Tissues, and Organisms, constructing a comprehensive knowledge graph. The database encompasses 7,145,790 substances and 252,486,973 bioactivity data points, offering a significantly more extensive repository of drug mechanism information than existing databases. This resource is valuable for various applications, including predicting drug efficacy and side effects, elucidating intermediate pathways in predictive models, and enhancing the drug development process.
Korea Advanced Institute Science and Technology(KAIST)
Department of Bio and Brain Engineering(BBE)
Synergistic Bioinformatics Laboratory
Myeong-Ha Hwang*, Daeho Lee*, Sechan Lee, Hyojin Son, Haangik Park and Gwan-Su Yi†
- Journal: IEEE Journal of Biomedical and Health Informatics
- Status: Under Review
- First Author(*): hmh929@kaist.ac.kr, daeho715@gmail.com
- Corresponding Author(†): gwansuyi@kaist.ac.kr
This work was supported by the BK-21 program through National Research Foundation of Korea (NRF) under Ministro of Education.