Environmental contaminants cause 16% of premature deaths worldwide, yet most chemicals in commercial use have never been assessed for their potential to cause harm to humans or ecosystems. Traditional toxicity testing methods that use live animals are too expensive, slow, and ethically concerning to assess chemical hazard at scale.

Drug discovery programs routinely screen thousands to millions of compounds in computational and cellular models for their potential to treat disease. What if we used this same technology to assess whether chemicals used in other sectors of society are dangerous? Scientists and government regulators are urgently working to implement high-throughput screening methods to prioritize resource-instensive testing for the most hazardous compounds.

A brute-force approach of testing every compound in all cell types and species is impractical. Instead, scientists must develop computational approaches that leverage prior knowledge of taxonomic relationships, cell types, and cell state to design efficient chemical screens and extrapolate observed toxicity across biological space.