DDI-GPT: Explainable Prediction of Drug-Drug Interactions using Large Language Models enhanced with Knowledge Graphs
DDI-GPT web server for knowledge-enhanced DDI exploration. 1. The user submits a pair of drugs, referenced by DrugBank ID, the input sentence is generated by incorporating drug-related biomedical entities from KG. The input is then set to a cluster running the prediction pipeline, where it is processed. 2. Once the interface prediction is complete, the DDI-GPT web server will be updated to visualize importance words generated by explanation module. 3. DDI-GPT web server allows users to explore the KGs that were used by DDI-GPT framework. as well as various drug-protein-drug interaction networks and drug-side-effects-drug interaction network. 4. The user can view shortest paths or two-hop paths between the combination of drugs.
- Online: https://pyvisddi-24u28afk4upfhpvclyujvs.streamlit.app/
- Local excution:
streamlit run Introduction.py