XenoSite is a web-based computational resource that predicts how small molecules may be transformed by liver enzymes and which resulting metabolites could pose toxicity risks. Designed to support drug metabolism and safety assessment, it offers models covering several important processes, including Phase I metabolism, molecular reactivity, epoxidation, quinone formation, N-dealkylation, and UGT-mediated conjugation. Its predictions can identify likely sites of metabolism or reactivity on a molecule, helping researchers examine potential metabolic pathways and reactive products during drug discovery and lead optimization.

The site’s models are based on machine-learning methods trained and evaluated using metabolic and chemical reactivity data. XenoSite is available free for academic and non-commercial use, while licenses provide secure commercial access and access to a programmatic API. The website is under active development, with additional models and features planned. It also provides a forum for questions, comments, and suggestions, and cites the scientific publications behind its prediction methods.