Knowledge Graph & Large Language Model-Enhanced Organ on a Chip Database (KLOCD)

KLOCD is a web platform for organ-on-a-chip research, drug discovery and toxicology. It consolidates data from public databases into eight sub-databases (drugs, chemicals, gene expression datasets, diseases, toxicants, literature, patents and organ-on-chip models) and adds AI tools built on its knowledge graph.

OCKG

The Organ-on-Chip Knowledge Graph (OCKG) links organ chips, diseases, genes and compounds. Version 1.0 has 76,540 nodes of 17 types and 2,315,561 relationships of 21 types. It supports two tools:

• Drug repurposing: link prediction and reinforcement learning over OCKG to predict drug-disease associations.
• Intelligent assistant: a retrieval-augmented question answering system that combines OCKG data with large language models.

Data Sources

The Help page lists these sources: PubMed, the FDA, PubChem, TogoWS, UniChem, DrugBank, ChEMBL, LiverTox, ClinicalTrials.gov, AACT, the CDC, Health Canada, the European Medicines Agency, CTD and the NCI Biospecimen Research Database. Its “GSA” gene expression datasets were retrieved from NCBI GEO, and the homepage also links Disease Ontology and openFDA. For literature that is not open access, KLOCD stores only titles, abstracts and links.

History

KLOCD succeeds the Organ-on-a-Chip Database (Ocdb), which a 2022 review from the same group described at the same address (http://www.organchip.cn/). Ocdb collected literature, patents, toxicological and pharmaceutical data, and was built on MySQL.

Access and Terms

KLOCD is served over HTTP only; HTTPS connections failed when checked on 2026-10-03. Some features require registration and login, and no bulk download or API is offered. No license is stated: the Help page says all data sources are open-access databases used according to their own terms. No public contact email was found; the About page names Professor Zhongze Gu as the team lead.