You are here:
ExpoPath: Identifying and Annotating Exposure Pathways from Chemical Co-occurrence Networks
Citation:
Zurek-Ost, M., K. Phillips, A. Williams, A. Edelman-Munoz, S. Handa, AND K. Isaacs. ExpoPath: Identifying and Annotating Exposure Pathways from Chemical Co-occurrence Networks. SOT, Salt Lake City, UT, March 10 - 14, 2024. https://doi.org/10.23645/epacomptox.25408519
Impact/Purpose:
This abstract describes a novel approach to understanding chemical co-occurrence using network analysis. This is an abstract for the 2024 SOT meeting.
Description:
Background and Purpose An ongoing and primary research endeavor of the US EPA includes the evaluation of chemical risk to environmental and human health as well as modes of traversal from commercial sources and industrial processes to environmental and ecological media (including receptors). Understanding relevant exposure pathways remains integral for the EPA’s commitment to chemical prioritization and regulation. Continued research into exposure pathways directly supports this commitment by uncovering new connections between source and media and identifying likely means of traversal used in pathway predictions. Methods This research adopts a comprehensive approach to the study of exposure pathways by applying network analysis methodologies to chemical co-occurrence networks as a means of empirically deriving exposure pathways based on relational observations of chemicals in sources and receptors. Source data from EPA’s Toxic Substances Control Act’s (TSCA) Chemical Data Reporting (CDR) rule, the EPA’s Chemicals and Products Database, the EPA’s internal Chemical Transformations Database (ChET), DrugBank, and US Food and Drug Administration databases, as well as media occurrence data from EPA’s Multimedia Monitoring Database (MMDB) were aggregated and used to generate the chemical co-occurrence network of chemicals occurring in at least one source and one medium. The Stanford Network Analysis Platform’s (SNAP) Communities from Edge Structure and Node Attributes (CESNA) tool was utilized to detect overlapping community structure of these chemicals. General co-occurrence between the chemicals accounted for the edge structure of the network while the “presence-in-source/media” incidence matrix was assigned as the node attributes preserving the media-specific patterns of co-occurrence deemed influential to the assignment of the overlapping communities. Results Results indicated over 20 unique overlapping communities of chemicals representing linkages among sources and media; examples of well-defined communities included chemicals co-occurring in 1) pesticides, surface water, and yard/landscape consumer products and 2) petrochemical manufacturing, consumer electronics, groundwater, and drinking water. These results demonstrate the usefulness of adopting overlapping communities as a means of representing empirically derived chemical exposure pathways. Logistic weights detail likelihood of community membership, suggest both intuitive and novel modes of traversal, and prove invaluable for annotating such pathways. Future efforts will aim to enhance the robustness and coverage of these analyses, including the incorporation of Non-Targeted Analysis (NTA) confirmations, use-cases for methods of edge prediction, as well as data management strategies related to consolidation of and selection procedures for media/receptor classification. Conclusions This project showcases the benefit of a network analysis approach to the study and annotation of exposure pathways derived from known, and predicted, chemical occurrences. Understanding the origin and fate of chemicals sharing common mechanisms of toxicity have the potential to inform chemical prioritization and mitigation strategies.
URLs/Downloads:
DOI: ExpoPath: Identifying and Annotating Exposure Pathways from Chemical Co-occurrence Networks
POSTER.PDF (PDF, NA pp, 897.697 KB, about PDF)