You are here:
Using EPA’s ChemExpo Tool to Build Updated Quantitative Structure-Use Relationship (QSUR) Models
Citation:
Phillips, K., V. Hull, S. Handa, AND K. Isaacs. Using EPA’s ChemExpo Tool to Build Updated Quantitative Structure-Use Relationship (QSUR) Models. ISES, Chicago, IL, August 27 - 31, 2023. https://doi.org/10.23645/epacomptox.24029985
Impact/Purpose:
If accepted, this abstract will lead to a presentation at the International Society of Exposure Science which will allow researchers from around the world to learn about our work on making exposure-relevant data more accessible to the public as well as how we can begin to fill gaps in those data with machine learning models.
Description:
There are thousands of chemical substances in need of prioritization for risk evaluation by the U.S. Environmental Protection Agency (EPA). While high-throughput toxicology methods can be utilized to obtain screening level hazard values for most of these chemicals, there are still gaps in data available to support analogous methods for developing screening level exposure values. Recently, the EPA's Office of Research and Development has developed a suite of tools to aid in advancing high-throughput exposure estimates. One tool in this suite is the Chemicals and Products Database (CPDat) which contains information on composition and use of chemicals in consumer and industrial settings. In addition to CPDat, EPA has developed an internal data management and curation tool, Factotum, that facilitates rapid collection, curation, and categorization of new sources into CPDat. Here, we demonstrate how CPDat (and the underlying Factotum) can aid in constructing new Quantitative Structure-Use Relationship (QSUR) models. These models have proven to be helpful in parameterizing high-throughput exposure models (including estimating key emissions to indoor and ambient environments), bolstering tentative identifications from non-targeted analyses, and identifying candidate chemical alternatives. Information required to build and validate these models can quickly be identified and downloaded via EPA’s new public-facing Chemical Exposure Knowledgebase (ChemExpo) application for exploring CPDat data. QSUR models for 44 different functional uses were constructed and validated with a training set of more than 10k chemical substances. Finally, we evaluate these models against public regulatory substance inventories.
URLs/Downloads:
DOI: Using EPA’s ChemExpo Tool to Build Updated Quantitative Structure-Use Relationship (QSUR) Models
POSTER.PDF (PDF, NA pp, 4021.501 KB, about PDF)