| Autor: |
Herkert NJ; Nicholas School of the Environment, Duke University, Durham, North Carolina 27710, United States., Getzinger GJ; School of Environmental Sustainability, Loyola University Chicago, Chicago, Illinois 60660, United States., Hoffman K; Nicholas School of the Environment, Duke University, Durham, North Carolina 27710, United States., Young AS; Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts 02115, United States., Allen JG; Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts 02115, United States., Levasseur JL; Nicholas School of the Environment, Duke University, Durham, North Carolina 27710, United States., Ferguson PL; Department of Civil and Environmental Engineering, Duke University, Durham, North Carolina 27708, United States., Stapleton HM; Nicholas School of the Environment, Duke University, Durham, North Carolina 27710, United States. |
| Abstrakt: |
Wristband personal samplers enable human exposure assessments for a diverse range of chemical contaminants and exposure settings with a previously unattainable scale and cost-effectiveness. Paired with nontargeted analyses, wristbands can provide important exposure monitoring data to expand our understanding of the environmental exposome. Here, a custom scripted suspect screening workflow was developed in the R programming language for feature selection and chemical annotations using gas chromatography-high-resolution mass spectrometry data acquired from the analysis of wristband samples collected from five different cohorts. The workflow includes blank subtraction, internal standard normalization, prediction of chemical uses in products, and feature annotation using multiple library search metrics and metadata from PubChem, among other functionalities. The workflow was developed and validated against 104 analytes identified by targeted analytical results in previously published reports of wristbands. A true positive rate of 62 and 48% in a quality control matrix and wristband samples, respectively, was observed for our optimum set of parameters. Feature analysis identified 458 features that were significantly higher on female-worn wristbands and only 21 features that were significantly higher on male-worn wristbands across all cohorts. Tentative identifications suggest that personal care products are a primary driver of the differences observed. |
