Science Inventory

20191007 - A public database supporting evidence-based exposomics (NCSOT)

Citation:

Sayre, R., J. Wambaugh, K. Phillips, A. Williams, AND C. Grulke. 20191007 - A public database supporting evidence-based exposomics (NCSOT). 2019 NCSOT Annual Meeting, Research Triangle Park, North Carolina, October 07, 2019. https://doi.org/10.23645/epacomptox.11877429

Impact/Purpose:

To support identification of likely sources of chemicals found in biological media through non-targeted/suspect screening analysis (SSA/NTA), this work-in-progress annotates chemicals with likely origin categories and adds empirically-validated substance relationships between chemicals and their in vivo transformation products to the CompTox Chemicals Dashboard.

Description:

To support identification of likely sources of chemicals found in biological media through non-targeted/suspect screening mass spectrometry analysis, our project adds substance relationships between chemicals and their transformation products to the CompTox Chemicals Dashboard. We propose five categories for substances found in biomonitoring samples: (1) endogenous human metabolome, (2a) exogenous nutrients, (2b) markers of exposure to exogenous nutrients, (3a) xenobiotics, and (3b) markers of exposure to xenobiotics. Some compounds, such as formaldehyde, can appear in more than one category. Our effort is unique in that we include only in vivo empirically observed relationships, rather than products formed in vitro, in different species, or predicted based on pathways. Restriction to observable transformations will also include products formed by biotic, abiotic, or complex processes that have not yet been identified as relevant pathways for toxicology. We have curated a set of over 10,000 observations of 3a/3b chemical pairs and contextual metadata (such as analytical method) from databases and literature. Databases were filtered to only include empirical pairs and literature mappings were identified with natural language processing and manually verified. Description of experiments for improving the machine learning workflow to identify and extract this data will be the focus of this poster. Substance relationship mappings for curated substances are visible in the CompTox Chemicals Dashboard as a publicly available resource for top-down exposomics, in which biomarker analytes measured are used to infer association with exposure agents. These mappings also allow for the development of exposure estimates based on dose levels demonstrated to yield a detectable amount of product, and analysis of toxicity based on a greater number of relevant species. This abstract does not necessarily represent the views or policies of the U.S. Environmental Protection Agency.

Record Details:

Record Type: DOCUMENT ( PRESENTATION/ POSTER)
Product Published Date: 10/07/2019
Record Last Revised: 02/20/2020
OMB Category: Other
Record ID: 348250