The TOXIN knowledge graph: supporting animal-free risk assessment of cosmetics.

Summary

This resource paper presents the TOXIN knowledge graph, an ontology-driven platform integrating SCCS cosmetic ingredient data, study reliability scoring, chemical identifiers, and QSAR predictions to support animal-free risk assessment. Populated with 88 ingredients, it flagged 53 compounds with liver toxicity signals and connected findings to adverse outcome pathways like hepatic cholestasis.

Key Findings

• Built an ontology-based knowledge graph integrating SCCS opinions (2009–2019), ToxRTool reliability, SMILES, and OECD QSAR Toolbox.
• Populated with 88 cosmetic ingredients and identified 53 compounds affecting at least one liver toxicity parameter in 90-day studies.
• Linked findings to adverse outcome pathways, exemplified by hepatic cholestasis for one compound.
• Provided interactive visualization and filtering to surface liver toxicity–related compounds for NGRA.

Clinical Implications

While not a clinical study, the tool can prioritize cosmetic ingredients for human-relevant in vitro testing, streamline hazard assessment, and reduce reliance on animal data, informing safer product development.

Limitations

• Current coverage limited to SCCS-derived data and primarily liver-focused endpoints.
• Signals require confirmation with human-relevant NAMs; exposure and dose-response integration are pending.

Future Directions

Expand ingredient coverage, incorporate exposure/dose-response and uncertainty quantification, and prospectively validate KG-derived hypotheses with human NAMs within NGRA workflows.