Skip to main content
NCBI home page
EPA Author Manuscripts logoLink to EPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jul 16.
Published in final edited form as: Anal Chem. 2025 Jul 16;97(29):15904–15912. doi: 10.1021/acs.analchem.5c02223

Prioritizing Chemical Candidates from Non-targeted Analysis Using Metadata, Spectral Similarity, and Hazard Scoring within INTERPRET NTA

Alex Chao 1,, Jeffrey M Minucci 2,, Troy M Ferland 3,4, E Tyler Carr 5,6, Greg Janesch 7,8, Safia Rizwan 9,10, Heather D Whitehead 11, Tommy Cathey 12, Shirley Pu 13,14, Laura D Brunelle 15,16, Angela L Batt 17, Jon R Sobus 18, Antony J Williams 19
PMCID: PMC13370747  NIHMSID: NIHMS2170860  PMID: 40670324

Abstract

New approach methodologies like non-targeted analysis (NTA) are increasingly relevant for identifying and monitoring emerging contaminants. Utilizing high-resolution mass spectrometry (HRMS), NTA methods can detect and annotate chemicals without prior knowledge. Yet, NTA results often have associated uncertainties owing, in part, to a lack of standardized methods. To address this challenge, the US Environmental Protection Agency (US EPA) developed the Interface for Processing, Reviewing, and Translating NTA data (“INTERPRET NTA”) to support NTA research within its Office of Research and Development (ORD). Previous work demonstrated INTERPRET NTA’s capabilities for reviewing and reporting NTA data quality. The current work highlights additional functionalities for retrieving and interpreting chemical results. INTERPRET NTA accesses chemical data from multiple US EPA resources, including (1) chemical metadata from US EPA’s Analytical Methods and Open Spectra (AMOS) database, (2) predicted spectra for ~1.2 million chemical substances within US EPA’s DSSTox database, and (3) hazard values from US EPA’s Cheminformatics Hazard Module (CHM). De facto water reuse study data with 77 known chemicals were used to demonstrate INTERPRET NTA functions and capabilities. Known chemicals showed higher values for metadata, MS2, and hazard scores in 99.0%, 80.5%, and 92.0% of cases, respectively, compared to false positives. Interactive visualizations within INTERPRET NTA facilitate the visual integration of chemical results, highlighting chemical candidates of greatest interest, and allowing review of underlying data. INTERPRET NTA is being prepared for public release, offering researchers a tool for defensible and efficient review and reporting of NTA study results.

Graphical Abstract

graphic file with name nihms-2170860-f0001.jpg

1. INTRODUCTION

Over the past half-century, US federal and state health agencies have researched the effects of chemical exposures on human and ecological health, implemented restrictions on exposures in cases of unacceptable risk, and developed monitoring programs with which to assess regulatory compliance. Under existing assessment paradigms, less than one percent of known chemicals have undergone rigorous testing and been listed on monitoring inventories for air, water, food, consumer products, or other regulated materials.1,2 Limited testing and monitoring reflects the extensive time and cost associated with traditional assessment protocols. To address this limitation, “new approach methodologies” (NAMs) have been developed, enabling provisional hazard characterization for thousands of analytes using in vitro testing and in silico modeling results.3–6 Exposure NAMs also exist, with efforts focused on modeling behaviors and generating monitoring data for many data-poor chemicals.7,8 The primary exposure monitoring NAM is “non-targeted analysis” (NTA), which uses high-resolution mass spectrometry (HRMS) to detect, identify, and quantify chemicals in studied samples without predefined targets. HRMS technologies have existed for decades, but more affordable benchtop instruments have only recently become accessible to federal and state environmental monitoring laboratories. Given this availability, NTA methods are poised to aid the characterization of emerging contaminants in the US and around the globe.

The general premise of an NTA experiment is that acquired HRMS data can help deduce the chemical contents of a given sample without having (1) prior knowledge of the sample’s chemical composition, or (2) interest in defined chemical targets within the sample. Results of NTA studies are intended to supplement those of traditional targeted methods, such as those outlined in numbered methods by the US Environmental Protection Agency (US EPA), and help prioritize emerging contaminants recommended for further safety assessment.9,10 While the development and utilization of NTA methods has accelerated in recent years, practices are not yet standardized, with hardware, software, databases, algorithms, and reporting strategies varying across institutions, laboratories within institutions, and even investigators within laboratories. This presents a considerable challenge for the NTA research community. On one hand, investigators should be encouraged to push experimental boundaries, developing new tools and methods that facilitate wider detectable space coverage and improved certainty in annotations and quantitative estimates. On the other hand, the NTA research field should consciously and collaboratively move toward analysis and reporting standardization to ensure transparency and reproducibility.11 This is especially true for regulatory organizations who may use NTA data to inform provisional decisions related to chemical safety.12–17

At the US EPA, NTA research is performed across the Office of Research and Development (ORD) within all four Centers. Research efforts support the identification of potential chemical stressors (including transformation and degradation products) in air, watersheds, drinking water, sediment and soil, marine and terrestrial biota, consumer products, human samples, and other matrices.18–26 They further support metabolic mapping of chemicals to improve understanding of perceived (either in vivo or in vitro) hazards and dose–response relationships.27,28 Finally, they have been designed to support rapid response efforts for environmental emergencies, including wildfires and unintended chemical releases.29,30 US EPA’s clients, which include program offices, regional offices, states, and tribes, cannot make the most informed and defensible decisions without confidence in ORD’s NTA data. Confidence is best ensured through clear communication of rigorous science that adheres to accepted standards and consensus best practices.9,11,31–34

In a move toward standardization, US EPA scientists have developed processes and tools to improve data quality review and harmonize reporting.33,35,36 Most notably, US EPA’s INTERPRET NTA (Interface for Processing, Reviewing, and Translating NTA Data; hereafter, “I-NTA”) application is available to all US EPA NTA researchers seeking a standardized data processing and reporting platform. In a recent article, we described the functionality of I-NTA with respect to automated quality assurance/quality control (QA/QC) review of preprocessed MS1 data (i.e., MS1 data that has undergone deconvolution, componentization, and alignment).35 This previous article provided great detail on methods, calculations, and visualizations used to support data postprocessing for quality documentation. The current article extends the I-NTA story, providing a description of application functionality for chemical annotation and prioritization.

Herein, using published NTA data from a de facto water reuse study, we demonstrate how I-NTA identifies chemical candidates from US EPA’s Distributed Structure-Searchable Toxicity (DSSTox) database of ~1.2 million chemical substances. We further show how I-NTA prioritizes DSSTox candidates via (1) metadata retrieval and scoring (using US EPA’s Analytical Methods and Open Spectra [AMOS] database);37 (2) MS2 spectral match scoring (using prepredicted in silico spectra available within AMOS); and (3) hazard/exposure data scoring (using measured and predicted hazard values, and predicted exposure values from US EPA’s Cheminformatics Modules). Our example demonstration on the de facto water reuse data set shows how agency NTA scientists can commonly access highly curated chemistry data, use copious amounts of Supporting Information to prioritize candidates for further review, and document all data processing steps that may affect agency and/or partner decisions.

2. EXPERIMENTAL SECTION

2.1. The INTERPRET NTA Application.

I-NTA is currently available as an internal tool for use by US EPA researchers and collaborators with approved access credentials; a broader public release is forthcoming. I-NTA is primarily a data processing and analysis engine with linkages to various Agency chemical databases that can support NTA chemical candidate identification and prioritization (Figure S1). DSSTox serves as the underlying chemical substance database for I-NTA and has been accessible to the public for almost a decade via the web-based application titled the “CompTox Chemicals Dashboard” (hereafter, the “Dashboard”), available at https://comptox.epa.gov/dashboard.38 I-NTA’s web interface is hosted as a module of the AMOS web application (https://hcd.rtpnc.epa.gov/#/amosapp/).

The AMOS web application was developed as (1) a content management system for analytical method documents mapped to chemical substances; and (2) a repository of spectra (from gas chromatography [GC] and liquid chromatography [LC] coupled mass spectrometers, and NMR, to date) compiled from public sources and user submissions.37 While AMOS was initially developed as a wholly separate application from I-NTA, integration was deemed appropriate and necessary to facilitate efficient utilization of both experimental spectra (~924k spectra associated with ~34k substances) and in silico spectra (generated using CFM-ID version 4.4.7 for electrospray [ESI] predictions and version 2.4.2 for electron ionization [EI] predictions) within agency NTA projects (as demonstrated to be highly beneficial in our previous works).39–42

I-NTA is a module of AMOS and allows for the upload and analysis of HRMS data files (Figure 1). At the time of publication, three separate workflows are supported for LC HRMS data: MS1, MS2, and Merge. For the MS1 workflow, data are uploaded via comma-separated value (.csv) templates with fields for feature mass, retention time (RT), and measured intensity across all samples. This workflow primarily focuses on two key components: the generation and review of QA/QC metrics, and generation of chemical results that encompass metadata, hazard, and exposure values retrieved from multiple US EPA databases.35 For the MS2 workflow, data are uploaded as mascot generic format (.mgf) or mass spectrometry peak (.msp) files, which can be exported from most HRMS software, and these spectral data are scored against in silico MS2 spectra generated from US EPA’s DSSTox database of chemicals. For the Merge workflow, users input the final results files from the MS1 and MS2 analyses to enable full data integration. Results from the relevant workflows are accessible via a URL containing the randomly generated job identification number assigned to the run. The MS1, MS2, and Merge analysis output consists of a compressed (.zip) directory containing either a .xlsx or .csv file of tabular results from each major processing step. The web interface also provides a series of interactive visualizations including decision trees, a data quality heatmap, run sequence plots, and an interactive chemical results scatterplot and bar charts for the MS1 workflow (an I-NTA User Guide, available: https://whitehead-heather.github.io/INTERPRETNTA/, provides details of interactive visualizations). Additional details regarding I-NTA and associated US EPA databases can be found in Section S1.1.

Figure 1.

Figure 1.

Open in a new tab

Overall workflow diagram of the INTERPRET NTA application for the MS1, MS2 and Merge workflows. Included are details for the input data and specific US EPA databases queried within each workflow, along with resulting outputs. Modules previously described for AMOS and INTERPRET NTA are highlighted in purple and beige, respectively. Modules newly introduced in this work are highlighted in green.

2.2. NTA Data Set for Proof-of-Concept Demonstration.

Study details and initial findings from a de facto water reuse study were first described in Brunelle et al.20 Briefly, polar organic chemical integrative samplers (POCIS) were employed at four locations during three separate time points along a surface water flow path in the Northeast U.S., and samples were analyzed with technical triplicates. The data used here are based on the subsequent I-NTA QA/QC demonstration work of Sobus et al. (2025) on this data set.35 Preprocessed MS1 data from Sobus et al. and available MS2 data were further processed via the I-NTA MS1 and MS2 workflows, respectively, prior to being merged. In the original work of Brunelle et al., a system suitability mixture (SSM), containing 63 method-amenable native (i.e., nonisotopically labeled) compounds, was spiked into a pooled QC sample and analyzed alongside study sample extracts. Examination of SSM chemicals by Sobus et al. enabled an extensive quality review. A summary of quality scores for SSM chemicals are provided in Table S1.35 Twenty-six chemicals were confirmed as confidence level 1 identifications by Brunelle et al. and included in the data set reported by Sobus et al. Twelve of these chemicals overlapped with the SSM chemical set. Thus, 77 unique chemicals (Table S2) are known to exist in study samples (including the spiked pooled QC sample). These chemicals are hereafter referred to as the known chemicals, and their associated features as the known chemical features. These known identities serve as true positives to demonstrate how chemical metadata, in silico spectra, and hazard values can support efficient chemical identification and prioritization in NTA studies. Additional details regarding the data set, data processing via I-NTA, and scoring of results are described in Section S1.2.

3. RESULTS AND DISCUSSION

3.1. Descriptive Statistics for Features, Candidates, and Substances.

The final occurrence matrix of the MS1 workflow (which contains only data passing all I-NTA QA/QC filters) contained 3331 features. All 77 known chemicals (Table S2) were identified in the final occurrence matrix. Sixty-five positive mode features and 22 negative mode features mapped to the 77 known chemicals. Thus, a total of 87 known chemical features exist in the MS1 workflow output. Data for these features are hereafter used for I-NTA workflow demonstration.

Mass searching the 3331 features against the DSSTox database returned 304,140 MS-Ready chemical structures, which mapped to 340,748 chemical substances. For the 87 known chemical features, 8096 chemical structures were retrieved, corresponding to 10,096 substances. Most feature searches returned between 0 and 100 candidate structures (Figure S2a,b). On average, approximately 90 incorrect candidate structures were retrieved for each known chemical feature.

3.2. Descriptive Statistics for Metadata.

Metadata have been demonstrated to assist in tentative chemical identification for NTA studies.43 Median metadata counts for correct candidates were in the tens-of-thousands for PubChem patents, PubChem articles, and PubMed articles; in the hundreds for PubChem sources and AMOS open spectra; and in the tens and ones for AMOS methods, AMOS fact sheets, and Dashboard water lists (Table S3 and Figure S3). Due to the variability in count numbers across different metadata sources, normalization was deemed necessary prior to combination into a total metadata score (Section S3.0).

Paired ratios for all known chemical features are shown in Figure S4, with percentages indicating how often the correct candidate for a given feature had higher, equivalent, or lower counts than corresponding incorrect candidates for each metadata field and the total metadata score. For seven of the eight metadata fields, the percentage of ratios greater than one ranged from 96.4% to 98.8%. AMOS fact sheets were the only field outside this range, with a value of 56.9% (likely caused by low absolute counts, as shown in Figure S3). The total metadata score had the highest percentage of ratios greater than one (99.0%), indicating value in combining disparate metadata counts (Figure S4).

3.3. Descriptive Statistics for MS2 Scores.

Twenty-three percent of all features (n = 782) and 98% of known chemical features (n = 85) had acquired MS2 data. MS2 database searching produced 73,009 candidates for all features (Figure S5a) and 9567 candidates for known chemical features (Figure S5b). Median MS2 quotient scores for correct and incorrect candidates of known chemical features were 0.87 and 0.51, respectively (Table S3 and Figure S3). Regarding ratios of MS2 scores between paired correct and incorrect candidates, 80.5%, 2.7%, and 16.8% of values were greater than, equal to, and less than one, respectively (Figure S4).

3.4. Descriptive Statistics for Hazard Scores.

All quality-adjusted hazard (QAH) scores and hazard completion scores based on the full Cheminformatics Hazard Module (CHM) query are shown in Figure S6. Median values for individual CHM end points are additionally given in Table S4. For known chemical features, median QAH scores of correct and incorrect candidates were 4.50 and 1.50, respectively, and hazard completeness scores were 0.25 and 0.05, respectively (Table S3 and Figure S3). Regarding ratios of QAH scores between paired correct and incorrect candidates, 92.0%, 4.7%, and 3.2% of values were greater than, equal to, and less than one, respectively (Figure S4).

3.5. Correlations between Metadata, MS2, and Hazard Scores.

Patent, article, and source counts were positively correlated with each other, and with AMOS spectra counts, with Spearman correlation values ranging from 0.32 to 0.68 (Figure S7). The total metadata score showed some concordance with this group, with Spearman correlation values of 0.34–0.66. AMOS method, AMOS fact sheet, and Dashboard water list counts were positively correlated with each other, with Spearman correlation values ranging from 0.37 to 0.42. Correlations were generally minor between metadata counts, MS2 scores, and hazard scores, indicating a degree of orthogonality between these data elements (Figure S7).

3.6. Aggregate Scoring and Visualization.

Multidimensional scatterplots (Figure 2) were used to further interrogate all features with acquired MS2 data (n = 782). These plots portray each feature based on total metadata score (x-axis), MS2 score (y-axis), and hazard score (marker color), with marker size proportional to occurrence percentage (i.e., the percentage of samples in which the feature occurred). Each plot is organized into four quadrants to help establish priorities for more detailed investigation (see Section 3.7). For known chemical features (Figure 2A), 85.9% (n = 73) of correct candidates were present in Q1 (the space occupied by candidates with the highest total metadata scores and MS2 scores), followed by 3.5% (n = 3) in Q2 (highest MS2 scores, lowest metadata scores), 0.0% (n = 0) in Q3 (lowest metadata and MS2 scores), and 10.6% (n = 9) in Q4 (highest metadata scores, lowest MS2 scores). For incorrect candidates of known chemical features (Figure 2B), only 0.1% were found in both Q1 (n = 8) and Q4 (n = 5), whereas 39.4% (n = 3739) and 60.4% (n = 5730) were found in Q2 and Q3, respectively. Based on these findings, candidates in Q1 are most likely to yield positive confirmations, whereas those in Q3 are least likely.

Figure 2.

Figure 2.

Open in a new tab

Scatterplots of total metadata scores, MS2 scores, hazard scores, and occurrence percentages for: (A) correct candidates of known chemical features, (B) incorrect candidates of known chemical features, (C) all candidates of unknown features, and (D) filtered candidates of unknown features (minimum 50% occurrence percentage and minimum hazard score of 4). Scatterplots are broken into priority quadrants Q1–Q4, with labels for absolute counts and percentages of candidates present in each quadrant.

For candidates of unknown features (Figure 2C), 0.5% (n = 309) were found in Q1. These candidates represent potential “low-hanging fruit” and are likely to yield the greatest number of confirmatory results upon additional analysis. Correct candidates can also be found in Q2, Q3, and Q4, but these quadrants represent areas of potentially higher effort and lesser gains. Prioritization across all four quadrants is possible by placing filters on hazard scores and/or occurrence percentages. Indeed, considering only candidates with at least 50% occurrence and a minimum hazard score of four (i.e., those with higher potential for exposure and risk) reduces the candidate pool in each quadrant by a factor of ~10 or more (Figure 2D).

3.7. Interactive Visualizations of Chemical Results.

Given diverse designs and goals of NTA studies, it is difficult to prescribe a one-size-fits-all approach for candidate interrogation. Rather than proposing a universal strategy, the I-NTA platform offers dynamic interactive visualizations that allow researchers to (1) control how chemical results are sorted, filtered, and displayed (thus aiding prioritization strategies in real-time), and (2) interrogate the underlying data of sorted, filtered, and displayed results based on visual interest. First, I-NTA’s interactive prioritization scatterplot (Figure 3) allows customization of up to four graphical dimensions (x-axis, y-axis, marker color, and marker size) in real-time, using feature information (e.g., median abundance and occurrence count) and candidate information (e.g., metadata, MS2, and hazard scores) to sort candidates for prioritization. Researchers may apply filters to feature and candidate information to exclude from the visualization candidates that do not meet user-defined criteria. The scatterplot supports interaction with candidate markers via hover-over and select-drag boxes that reveal detailed information on selected candidates and their associated features. Clicking on a candidate marker highlights all candidates for the associated feature within the scatterplot. A comprehensive description of all visualization functions of the interactive prioritization scatterplot is given in Section S5.1.

Figure 3.

Figure 3.

Open in a new tab

I-NTA interactive prioritization scatterplot of feature and candidate data. Four separate dimensions (x-axis, y-axis, marker color, marker size) are assignable to various data fields (MS2, metadata, and hazard scores; feature median abundance and occurrence count). Feature and candidate data are revealed on mousing over a given candidate. Associated candidates for a given feature are highlighted when a feature is selected.

Once features of interest are identified using the interactive prioritization scatterplot, the underlying candidates and their associated data may be further interrogated via interactive candidate bar charts (Figure 4) and tables. The dynamic bar charts provide a comprehensive view of the metadata, MS2, and hazard scores of all candidates for a selected feature. In the metadata bar chart (left panel), contributions from each metadata field are color-coded, and hovering over each sub-bar reveals metadata counts for the corresponding field. The MS2 bar chart (center panel) shows the normalized quotient MS2 score for each feature candidate. Finally, the hazard bar chart (right panel) visually displays both QAH scores and hazard completeness scores via bar length and bar shading, respectively. Each bar chart is independently sortable and mousing over individual candidates highlights their position within each chart. Clicking on candidates reveals their structure as well a direct link to their Dashboard chemicals page. An interactive table beneath the bar charts shows the underlying data used to generate each chart, allowing the researcher to see the original source data, while also allowing filtering and sorting of each bar chart by any specific metadata field. A comprehensive description of visualization functions of the interactive candidate bar charts is provided in Section S5.2.

Figure 4.

Figure 4.

Open in a new tab

I-NTA interactive candidate bar charts. For a selected feature, bar charts are shown for metadata, MS2, and hazard scores for associated candidates. Clicking on a candidate bar reveals its location and associated rank in each sorted bar chart and displays a clickable structure image in the bottom right corner that links to the CompTox Chemicals Dashboard page for the chemical candidate. A table containing the underlying data for each candidate is also displayed in the interactive visualization (shown in Section S5.2).

4. CONCLUSIONS AND PERSPECTIVES

NTA research is becoming increasingly vital to detect, identify, and characterize the chemicals in the environment that may adversely impact human and/or ecological health. Characterizing the “exposome”, or the totality of all exposures experienced over a lifetime, benefits from the development and use of software tools that enable efficient, reproducible, and transparent exposure characterization.34,44–48 The US EPA’s I-NTA application was introduced as a resource to meet this goal, with previously described functionalities covering NTA data quality review and reporting.35 The current work expands the description of I-NTA functionality with focus on the use of metadata, predicted spectra, and hazard values to aid NTA data interpretation.

An inherent bias exists in the current work due to well-known SSM and reference library chemicals being chosen as the focus of I-NTA demonstration. Specifically, because of their known relevance to drinking water exposure and human health, the selected chemicals are somewhat expected to stand out based on available metadata and hazard scores. The data resources and tools demonstrated herein are obviously not as well-suited for the identification of true unknowns, which are not captured in DSSTox or any other database. This work is therefore not meant to comment on the performance of I-NTA for identifying all experimental features, but rather to demonstrate the functionality of I-NTA for efficiently characterizing “known unknowns”. In light of our findings on a well-curated NTA data set, we have provided a clear roadmap for I-NTA application in future studies.

The I-NTA application is continuously evolving to better address the scientific needs of US EPA’s NTA research partners, which encompasses federal, state, and regional offices/laboratories. The inclusion of additional metadata, and the optimization of existing metadata selections, will continue as ORD develops new databases, prediction models, and tools of relevance to NTA. One future enhancement will include the incorporation of spectral scoring using AMOS experimental spectra. At present, only scoring against in silico spectra is utilized; the growing library of experimental spectra in AMOS (~900k spectra associated with ~34k compounds) is likely to be of high value within the MS2 and Merge workflows. Integration of a quantitative NTA (qNTA) workflow is also underway in I-NTA. This future workflow will allow concentration and uncertainty estimation for NTA features, with quality review outputs conforming to standardized qNTA performance metrics.49–51

I-NTA is currently available to US EPA researchers and partners only, with plans for public release in the future. I-NTA is, first and foremost, meant to standardize NTA research practices across ORD and directly support collaborative NTA projects at federal, state, regional, and tribal levels. Adoption of I-NTA improves the efficiency, transparency, and reproducibility of Agency NTA research activities, and thus, the defensibility of study outcomes. Armed with a tool that automates NTA quality review and reporting (in accordance with international best practices) and facilitates rapid data interpretation (using the most current and highly curated chemistry data), more research partners can perform defensible NTA research, thus enabling wider exposomics characterizations, and more timely provisional assessments of potential risk.

Supplementary Material

Supplement1
Supplement2

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.5c02223.

Supplementary text and figures (PDF)

Supplementary tables (XLSX)

ACKNOWLEDGMENTS

The authors thank Elin Ulrich and Michael DeVito for their helpful reviews; and Susan Glassmeyer, Marc Mills, David Alvarez, Edward Furlong, and Dana Kolpin for their roles in the de facto water reuse study.

Funding

This work was funded by the US EPA’s Office of Research and Development. This research was supported, in part, by a US EPA Research Participation Program, administered by the Oak Ridge Institute for Science and Education (ORISE), through an interagency agreement between the US Department of Energy (DOE) and the US EPA. ORISE is managed by Oak Ridge Associated Universities (ORAU) under DOE contract number DE-SC0014664.

Footnotes

The authors declare no competing financial interest.

The views expressed in this paper are those of the authors and do not necessarily represent the views or the policies of the US EPA. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government. This document has been reviewed in accordance with US EPA policy and is approved for publication.

Complete contact information is available at: https://pubs.acs.org/10.1021/acs.analchem.5c02223

Contributor Information

Alex Chao, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States.

Jeffrey M. Minucci, United States Environmental Protection Agency, Office of Research and Development, Center for Public Health and Environmental Assessment, Research Triangle Park, North Carolina 27711, United States.

Troy M. Ferland, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States Oak Ridge Institute for Science and Education (ORISE) Participant, Oak Ridge, Tennessee 37831, United States.

E. Tyler Carr, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States; Oak Ridge Affiliated Universities (ORAU) Student Services Contractor, Research Triangle Park, North Carolina 27711, United States.

Greg Janesch, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States; Oak Ridge Affiliated Universities (ORAU) Student Services Contractor, Research Triangle Park, North Carolina 27711, United States.

Safia Rizwan, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States; Oak Ridge Affiliated Universities (ORAU) Student Services Contractor, Research Triangle Park, North Carolina 27711, United States.

Heather D. Whitehead, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States

Tommy Cathey, General Dynamics Information Technology, Research Triangle Park, North Carolina 27711, United States.

Shirley Pu, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States; Oak Ridge Institute for Science and Education (ORISE) Participant, Oak Ridge, Tennessee 37831, United States.

Laura D. Brunelle, United States Environmental Protection Agency, Office of Research and Development, Center for Environmental Solutions and Emergency Response, Cincinnati, Ohio 45268, United States Oak Ridge Institute for Science and Education (ORISE) Participant, Oak Ridge, Tennessee 37831, United States.

Angela L. Batt, United States Environmental Protection Agency, Office of Research and Development, Center for Environmental Solutions and Emergency Response, Cincinnati, Ohio 45268, United States

Jon R. Sobus, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States

Antony J. Williams, United States Environmental Protection Agency, Office of Research and Development, Center for Computational Toxicology and Exposure, Research Triangle Park, North Carolina 27711, United States

REFERENCES

  • (1).Isaacs KK; Egeghy P; Dionisio KL; Phillips KA; Zidek A; Ring C; Sobus JR; Ulrich EM; Wetmore BA; Williams AJ; Wambaugh JF J. Exposure Sci. Environ. Epidemiol. 2022, 32 (6), 820–832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (2).Egeghy PP; Judson R; Gangwal S; Mosher S; Smith D; Vail J; Cohen Hubal EA Sci. Total Environ. 2012, 414, 159–166. [DOI] [PubMed] [Google Scholar]
  • (3).Environmental Protection Agency US. A proof-of-concept Case Study Integrating Publicly Available Information to Screen Candidates for Chemical Prioritization; U.S. Environmental Protection Agency: Washington, DC, 2021. [PubMed] [Google Scholar]
  • (4).Patlewicz G; Williams AJ; Adams M; Shah I; Paul-Friedman K Chem. Res. Toxicol. 2025, 38 (1), 129–144. [DOI] [PubMed] [Google Scholar]
  • (5).Westmoreland C; Bender HJ; Doe JE; Jacobs MN; Kass GEN; Madia F; Mahony C; Manou I; Maxwell G; Prieto P; Roggeband R; Sobanski T; Schütte K; Worth; Zvonar Z; Cronin MTD. Regul. Toxicol. Pharmacol. 2022, 135, 105261. [DOI] [PubMed] [Google Scholar]
  • (6).Parish ST; Aschner M; Casey W; Corvaro M; Embry MR; Fitzpatrick S; Kidd D; Kleinstreuer NC; Lima BS; Settivari RS; Wolf DC; Yamazaki D; Boobis A Regul. Toxicol. Pharmacol. 2020, 112, 104592. [DOI] [PubMed] [Google Scholar]
  • (7).Paul Friedman K; Thomas RS; Wambaugh JF; Harrill JA; Judson RS; Shafer TJ; Williams AJ; Lee J-YJ; Loo L-H; Gagné M; Long AS; Barton-Maclaren TS; Whelan M; Bouhifd M; Rasenberg M; Simanainen U; Sobanski T Toxicol. Sci. 2025, 205, 74–105. [DOI] [PubMed] [Google Scholar]
  • (8).Wambaugh JF; Bare JC; Carignan CC; Dionisio KL; Dodson RE; Jolliet O; Liu X; Meyer DE; Newton SR; Phillips KA; Price PS; Ring CL; Shin H-M; Sobus JR; Tal T; Ulrich EM; Vallero DA; Wetmore BA; Isaacs KK Curr. Opin. Toxicol. 2019, 15, 76–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (9).Sobus JR; Wambaugh JF; Isaacs KK; Williams AJ; McEachran AD; Richard AM; Grulke CM; Ulrich EM; Rager JE; Strynar MJ; Newton SR J. Exposure Sci. Environ. Epidemiol. 2018, 28 (5), 411–426. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (10).Buckley TJ; Egeghy PP; Isaacs K; Richard AM; Ring C; Sayre RR; Sobus JR; Thomas RS; Ulrich EM; Wambaugh JF; Williams AJ Environ. Int. 2023, 178, 108097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (11).Nason SL; McCord J; Feng Y-L; Sobus JR; Fisher CM; Marfil-Vega R; Phillips AL; Johnson G; Sloop J; Bayen S; Mutlu E; Batt AL; Nahan K Anal. Chem. 2025, 97 (5), 2567–2578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (12).van der Zalm AJ; Barroso J; Browne P; Casey W; Gordon J; Henry TR; Kleinstreuer NC; Lowit AB; Perron M; Clippinger AJ Arch. Toxicol. 2022, 96 (11), 2865–2879. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (13).Minnesota Pollution Control Agency. Draft NPDES and SDS Permit to 3M Chemical Operations’ Cottage Grove facility; Minnesota Pollution Control Agency, Online, 2024. https://shorturl.at/AiOFm (accessed April 29, 2025). [Google Scholar]
  • (14).State of New Jersey Office of the Attorney General Attorney General Platkin and DEP Commissioner Latourette Announce Proposed Settlement with Solvay Polymers Regarding “Forever Chemical” Pollution in and Around West Deptford; State of New Jersey Office of the Attorney General, 2023, Online. [Google Scholar]
  • (15).Chemours Consent Order; North Carolina Department of Environmental Quality, 2019. https://www.deq.nc.gov/news/key-issues/genx-investigation/chemours-consent-order (accessed 2025 July 15th). [Google Scholar]
  • (16).The Chemours Company. PFAS Non-targeted Analysis and Methods Interim Report: Process and Non-process Wastewater and Stormwater; The Chemours Company, 2020, Online. [Google Scholar]
  • (17).New Jersey Department of Environmental Protection New Interim Specific Ground Water Quality Criterion Established for Chloroperfluoropolyether Carboxylates (ClPFPECAs); New Jersey Department of Environmental Protection, 2022, Online. [Google Scholar]
  • (18).Rager JE; Strynar MJ; Liang S; McMahen RL; Richard AM; Grulke CM; Wambaugh JF; Isaacs KK; Judson R; Williams AJ; Sobus JR Environ. Int. 2016, 88, 269–280. [DOI] [PubMed] [Google Scholar]
  • (19).Panagopoulos Abrahamsson D; Sobus JR; Ulrich EM; Isaacs K; Moschet C; Young TM; Bennett DH; Tulve NS J. Exposure Sci. Environ. Epidemiol. 2021, 31 (1), 70–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (20).Brunelle LD; Batt AL; Chao A; Glassmeyer ST; Quinete N; Alvarez DA; Kolpin DW; Furlong ET; Mills MA; Aga DS Environ. Sci. Technol. 2024, 58 (5), 2468–2478. [DOI] [PubMed] [Google Scholar]
  • (21).Newton SR; McMahen RL; Sobus JR; Mansouri K; Williams AJ; McEachran AD; Strynar MJ Environ. Pollut. 2018, 234, 297–306. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (22).Lowe CN; Phillips KA; Favela KA; Yau AY; Wambaugh JF; Sobus JR; Williams AJ; Pfirrman AJ; Isaacs KK Environ. Sci. Technol. 2021, 55 (16), 11375–11387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (23).Phillips KA; Yau A; Favela KA; Isaacs KK; McEachran A; Grulke C; Richard AM; Williams AJ; Sobus JR; Thomas RS; Wambaugh JF Environ. Sci. Technol. 2018, 52 (5), 3125–3135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (24).Chao A; Grossman J; Carberry C; Lai Y; Williams AJ; Minucci JM; Thomas Purucker S; Szilagyi J; Lu K; Boggess K; Fry RC; Sobus JR; Rager JE Environ. Int. 2022, 167, 107385. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (25).Phillips KA; Chao A; Church RL; Favela K; Garantziotis S; Isaacs KK; Meyer B; Rice A; Sayre R; Wetmore BA; Yau A; Wambaugh JF Environ. Sci. Technol. 2024, 58 (4), 1802–1812. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (26).Bangma J; Pu S; Robuck A; Boettger J; Guillette T; McCord J; Rock KD; Sobus J; Jackson TW; Belcher SM Chemosphere 2024, 363, 142898. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (27).Bell SM; Chang X; Wambaugh JF; Allen DG; Bartels M; Brouwer KLR; Casey WM; Choksi N; Ferguson SS; Fraczkiewicz G; Jarabek AM; Ke A; Lumen A; Lynn SG; Paini A; Price PS; Ring C; Simon TW; Sipes NS; Sprankle CS; Strickland J; Troutman J; Wetmore BA; Kleinstreuer NC Toxicol. in Vitro 2018, 47, 213–227. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (28).Boyce M; Favela KA; Bonzo JA; Chao A; Lizarraga LE; Moody LR; Owens EO; Patlewicz G; Shah I; Sobus JR; Thomas RS; Williams AJ; Yau A; Wambaugh JF Front. Toxicol. 2023, 5, 1051483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (29).Sloop JT; Chao A; Gundersen J; Phillips AL; Sobus JR; Ulrich EM; Williams AJ; Newton SR Environ. Sci. Technol. 2023, 57 (8), 3075–3084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (30).Phillips AL; Williams AJ; Sobus JR; Ulrich EM; Gundersen J; Langlois-Miller C; Newton SR Environ. Toxicol. Chem. 2022, 41 (5), 1117–1130. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (31).Kutil Z; Novotna K; Cermakova L; Pivokonsky M Sci. Total Environ. 2024, 908, 168367. [DOI] [PubMed] [Google Scholar]
  • (32).Fisher CM; Peter KT; Newton SR; Schaub AJ; Sobus JR Anal. Bioanal. Chem. 2022, 414 (22), 6455–6471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (33).Peter KT; Phillips AL; Knolhoff AM; Gardinali PR; Manzano CA; Miller KE; Pristner M; Sabourin L; Sumarah MW; Warth B; Sobus JR Anal. Chem. 2021, 93 (41), 13870–13879. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (34).Lennon S; Chaker J; Price EJ; Hollender J; Huber C; Schulze T; Ahrens L; Béen F; Creusot N; Debrauwer L; Dervilly G; Gabriel C; Guérin T; Habchi B; Jamin EL; Klánová J; Kosjek T; Le Bizec B; Meijer J; Mol H; Nijssen R; Oberacher H; Papaioannou N; Parinet J; Sarigiannis D; Stravs MA; Tkalec Ž; Schymanski EL; Lamoree M; Antignac J-P; David A TrAC, Trends Anal. Chem. 2024, 174, 117674. [Google Scholar]
  • (35).Sobus JR; Sayre-Smith NA; Chao A; Ferland TM; Minucci JM; Carr ET; Brunelle LD; Batt AL; Whitehead HD; Cathey T; Boyce M; Ulrich EM; McCord JP; Williams AJ Anal. Bioanal. Chem. 2025, 417, 1897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (36).Phillips AL; Peter KT; Sobus JR; Fisher CM; Manzano CA; McEachran AD; Williams AJ; Knolhoff AM; Ulrich EM J. Exposure Sci. Environ. Epidemiol. 2023, 33 (4), 501–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (37).Janesch G; Carr ET; Sivasupramaniam S; Charest N; Tkachenko V; Williams AJJ Visualized Exp. 2025, 220, No. e68194. [DOI] [PubMed] [Google Scholar]
  • (38).Williams AJ; Grulke CM; Edwards J; McEachran AD; Mansouri K; Baker NC; Patlewicz G; Shah I; Wambaugh JF; Judson RS; Richard AM J. Cheminf. 2017, 9 (1), 61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (39).Allen F; Pon A; Wilson M; Greiner R; Wishart D Nucleic Acids Res. 2014, 42 (W1), W94–W99. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (40).McEachran AD; Balabin I; Cathey T; Transue TR; Al-Ghoul H; Grulke C; Sobus JR; Williams AJ Sci. Data 2019, 6 (1), 141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (41).Chao A; Al-Ghoul H; McEachran AD; Balabin I; Transue T; Cathey T; Grossman JN; Singh RR; Ulrich EM; Williams AJ; Sobus JR Anal. Bioanal. Chem. 2020, 412 (6), 1303–1315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (42).Wang F; Liigand J; Tian S; Arndt D; Greiner R; Wishart DS Anal. Chem. 2021, 93 (34), 11692–11700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (43).McEachran AD; Sobus JR; Williams AJ Anal. Bioanal. Chem. 2017, 409 (7), 1729–1735. [DOI] [PubMed] [Google Scholar]
  • (44).Hollender J; Schymanski EL; Ahrens L; Alygizakis N; Béen F; Bijlsma L; Brunner AM; Celma A; Fildier A; Fu Q; Gago-Ferrero P; Gil-Solsona R; Haglund P; Hansen M; Kaserzon S; Kruve A; Lamoree M; Margoum C; Meijer J; Merel S; Rauert C; Rostkowski P; Samanipour S; Schulze B; Schulze T; Singh RR; Slobodnik J; Steininger-Mairinger T; Thomaidis NS; Togola A; Vorkamp K; Vulliet E; Zhu L; Krauss M Environ. Sci. Eur. 2023, 35 (1), 75. [Google Scholar]
  • (45).Kirwan JA; Gika H; Beger RD; Bearden D; Dunn WB; Goodacre R; Theodoridis G; Witting M; Yu L-R; Wilson ID; Metabolomics Quality Assurance and Quality Control Consortium (mQACC). Metabolomics 2022, 18 (9), 70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (46).Viant MR; Ebbels TMD; Beger RD; Ekman DR; Epps DJT; Kamp H; Leonards PEG; Loizou GD; MacRae JI; van Ravenzwaay B; Rocca-Serra P; Salek RM; Walk T; Weber RJM. Nat. Commun. 2019, 10 (1), 3041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (47).Vermeulen R; Schymanski EL; Barabási A-L; Miller GW Science 2020, 367 (6476), 392–396. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (48).Wild CP Cancer Epidemiol., Biomarkers Prev. 2005, 14 (8), 1847–1850. [DOI] [PubMed] [Google Scholar]
  • (49).Pu S; McCord JP; Bangma J; Sobus JR Anal. Bioanal. Chem. 2024, 416 (5), 1249–1267. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (50).Groff LC; Grossman JN; Kruve A; Minucci JM; Lowe CN; McCord JP; Kapraun DF; Phillips KA; Purucker ST; Chao A; Ring CL; Williams AJ; Sobus JR Anal. Bioanal. Chem. 2022, 414 (17), 4919–4933. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (51).Pu S; McCord JP; Dickman RA; Sayre-Smith NA; Sepman H; Kruve A; Aga DS; Sobus JR Anal. Bioanal. Chem. 2025, 417, 2097. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement1
NIHMS2170860-supplement-Supplement1.pdf (3.2MB, pdf)
Supplement2
NIHMS2170860-supplement-Supplement2.xlsx (30.1KB, xlsx)

RESOURCES