Cross-Medicine Knowledge Graph for Type 2 Diabetes Drug Discovery

A knowledge graph constructed in 2026 integrates multi-source biomedical data to elucidate complex disease mechanisms. It contains 15 entity types (245,235 entities) and 52 relation types (7,155,373 triples), covering 709 core Type 2 Diabetes genes. The dataset was created by Zekun Zhou and is available under a CC-BY-4.0 license.

Use Cases

• Prioritizing potential drug candidates based on a unified path scoring framework incorporating graph embeddings and rule-based reasoning.
• Analyzing multi-target disease mechanisms, such as insulin signaling sensitization and inflammatory regulation, using the integrated biomedical relations.
• Benchmarking graph embedding models (TransE, DistMult, ComplEx, RotatE) for link prediction tasks on biomedical knowledge graphs.
• Exploring relationships between traditional Chinese medicine herbs and modern medicine compounds for Type 2 Diabetes treatment.

Strengths

• Integrates data from five distinct biomedical sources (Hetionet, SymMap, TCMBank, STRING, TTD).
• Contains a large-scale structure with over 245,000 entities and over 7 million relation triples.
• Focuses on a specific, complex disease (Type 2 Diabetes Mellitus) with coverage of 709 core genes.
• Includes a detailed methodological framework for interpretable, high-precision candidate prioritization.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for certain tasks.
• The dataset is very small (9.5 KB), suggesting it likely contains only a list or summary rather than the full graph data.

Provenance

Source

Integrated from Hetionet, SymMap, TCMBank, STRING, and TTD.

Collection Method

Constructed using Jaccard and overlap-based fusion strategies for entity alignment and relation consolidation.

Freshness

Last updated 2026-05-13 17:33:17; freshness should be verified.