DeepCyte launched DeeTox Atlas, a single-cell metabolomic reference dataset designed to map drug toxicity mechanisms with foundation-model scalability. The atlas combines two independent single-cell perturbation studies spanning ~100 toxicant compounds, ~300,000 cells, and ~500 metabolites per cell, with a curated toxicity hierarchy anchored to established adverse outcome pathways. DeepCyte says the foundation model can predict toxicity mechanisms for compounds not previously measured and that single-cell resolution can expose patterns in small subpopulations that bulk approaches may miss. The company also expects early enterprise pilots to begin with global pharma partners. Overall, the release positions DeeTox Atlas as a practical safety-enabling dataset intended to shift toxicity evaluation earlier in discovery rather than treating safety liabilities as late-stage surprises.