You are here:
Cheminformatics Analysis of EPA ToxCast Chemical Libraries to Identify Domains of Applicability for Predictive Toxicity Models and Prioritize Compounds for Toxicity Testing
Zhang, L., H. Zhu, I. Rusyn, R. JUDSON, D. J. DIX, K. A. HOUCK, M. T. MARTIN, A. M. RICHARD, R. J. KAVLOCK, AND A. Tropsha. Cheminformatics Analysis of EPA ToxCast Chemical Libraries to Identify Domains of Applicability for Predictive Toxicity Models and Prioritize Compounds for Toxicity Testing. Presented at Society of Toxicology Annual Meeting, Baltimore, MD, March 15 - 19, 2009.
We have used cheminformatics approaches including clustering, data visualization, and QSAR to develop models for EPA-320 that could help prioritizing EPA-10K validation chemicals. Both chemical descriptors, as well as calculated physicochemical properties have been used. Compounds from EPA-10K are prioritized based on their similarity to EPA-320 using different similarity metrics, with similarity thresholds defining the domain of applicability for the predictive models built for EPA-320 set. In addition, prioritized lists of compounds of increasing dissimilarity from the EPA-320 have been produced, to test the ability of the EPA-320 models to extrapolate. These lists are further prioritized based on availability of toxicity data in the EPA ACToR database (actor.epa.gov/actor).
An important goal of toxicology research is the development of robust methods that use in vitro and chemical structure information to predict in vivo toxicity endpoints. The US EPA ToxCast program is addressing this goal using ~600 in vitro assays to create bioactivity profiles on a set of 320 compounds, mostly pesticide actives, that have well characterized in vivo toxicity. These 320 compounds (EPA-320 set evaluated in Phase I of ToxCast) are a subset of a much larger set of ~10,000 candidates that are of interest to the EPA (called here EPA-10K). Predictive models of in vivo toxicity are being constructed from the in vitro assay data on the EPA-320 chemical set. These models require validation on additional chemicals prior to wide acceptance, and this will be carried out by evaluating compounds from EPA-10K in Phase II of ToxCast.