You are here:
New CSRML-based features to categorize and fingerprint PFAS structure lists for cheminformatics analysis and read-across
Citation:
Richard, A., R. Lougee, G. Patlewicz, Chris Grulke, A. Williams, C. Yang, J. Rathman, AND T. Magdziarz. New CSRML-based features to categorize and fingerprint PFAS structure lists for cheminformatics analysis and read-across. 19th International Workshop on (Quantitative) Structure-Activity Relationships in Environmental and Health Sciences (QSAR2021), Virtual, NC, June 07 - 10, 2021. https://doi.org/10.23645/epacomptox.15105216
Impact/Purpose:
Presentation to the 19th International Workshop on (Quantitative) Structure-Activity Relationships in Environmental and Health Sciences (QSAR2021), June 2021. Studies under this output will be designed to address data gaps associated with PFAS and to investigate whether the structural categories that had been defined to select the ~150 PFAS are supported by the NAM data being generated by HTT. The insights gained will be used to refine a harmonized scheme of categorizing PFAS. For PFAS with legacy in vivo data, the concordance of the NAM data relative to the in vivo data will also be evaluated in order to anchor what in vivo toxicity predictions can be made within the PFAS categories and with what level of scientific confidence. The results of these efforts will be of direct benefit to program and regional offices as well as the greater scientific community.
Description:
Per- and polyfluoroalkyl substances (PFAS) are of high interest due to widespread production, environmental persistence, and potential adverse ecological and health impacts. In 2019, the Organization for Economic Development (OECD) published a list of more than 4500 PFAS with potential environmental occurrence. EPA’s CompTox Chemicals Dashboard hosts a list with over 8000 curated structures spanning several public PFAS lists, inclusive of OECD’s. Studies to-date have focused on the health effects of a small number of PFAS compounds, such as PFOA and PFOS, whereas little is known about the health effects of the majority of PFAS and their byproducts. Means for profiling the PFAS chemical structure space are needed to support modeling and structure-based categorization efforts. However, publicly available molecular fingerprinting methods are ill-suited to effectively capturing PFAS-defining features and structure diversity. Similarly, PFAS chemical category terms are limited to simple categories (e.g., perfluorocarboxylic acids) and often lack clear structure definition. Building on publicly available ToxPrint (TxP) features coded in CSRML (Chemical Subgraphs and Reactions Markup Language), we developed 34 new TxP_PFAS features to serve as PFAS structural categories and more than 130 fingerprints to capture unique aspects of PFAS structures, such as perfluoro chains, polyfluoro substructures, fluorinated rings, and various perfluoro branching patterns, as well as functional groups. These CSRML PFAS categories and features can be used with the public Chemotyper, provide comprehensive coverage of available PFAS lists, and are being used to profile and describe PFAS chemical lists currently undergoing testing within EPA. Abstract does not reflect EPA policy.
URLs/Downloads:
DOI: New CSRML-based features to categorize and fingerprint PFAS structure lists for cheminformatics analysis and read-across
QSAR2021_POSTER#263_PFAS_ARICHARD_V2.PDF (PDF, NA pp, 1841.881 KB, about PDF)