STopTox: A machine learning app as an alternative to animal testing for acute systemic and topical toxicity

Acute toxicity tests are used to identify hazard potentials appearing after short exposure times. Despite some progress in reducing, refining, and replacing animal tests1, several types of acute toxicity in humans are demanded to be evaluated by the in vivo assays historically termed as “6-pack” (acute oral toxicity, acute dermal toxicity, acute inhalation toxicity, skin irritation and corrosion, eye irritation and corrosion, skin sensitization). We have developed a new in silico tool for the rapid identification of potential acute toxicants. We have compiled, curated, and integrated the largest publicly available datasets and built QSAR models for these six endpoints. All models were developed and validated according to the OECD QSAR principles2. Metrics of external model accuracy were evaluated. Models were made available within our publicly accessible STopTox web portal (http://stoptox.labmol.com.br/). These models can be employed by users to identify both putative toxicants and non-toxicants in chemical libraries of their interest.