MolSafeKG
Introduction
MolSafeKG is a heterogeneous knowledge graph designed for molecular safety assessment. It integrates 83,925 hazardous molecules from authoritative sources such as ECHA and ChEMBL, covering 8 categories of drug toxicity (e.g., carcinogenicity, hepatotoxicity) and 68 items across 3 GHS hazard classes (physical, health, and environmental). It is further linked to 149 functional groups and 434 structural alerts, forming a graph with more than 1.95 million triples. MolSafeKG supports explainable safety evaluation through similarity-based retrieval combined with LLM reasoning, enabling the detection of potential toxicity or hazardous properties in AI-generated molecules. It provides systematic safety-governance capabilities for molecular generation models, particularly suited for high-risk scenarios—such as drug discovery—where hallucination prevention and risk tracing are essential.
Chemistry
Domain