You are here:
Application of Functional Use Predictions to Aid in Structure Identification of Chemicals in House Dust
Citation:
Phillips, K., A. McEachran, J. Sobus, AND K. Isaacs. Application of Functional Use Predictions to Aid in Structure Identification of Chemicals in House Dust. American Chemical Society Spring Meeting, San Francisco, CA, April 02 - 06, 2017.
Impact/Purpose:
Humans are potentially exposed to thousands of anthropogenic chemicals in commerce. Recent work has shown that the bulk of this exposure may occur in near-field indoor environments (e.g., home, school, work, etc.). Advances in suspect screening analyses (SSA) now allow an improved understanding of the chemicals present in these environments. However, due to the nature of suspect screening techniques, investigators are often left with chemical formula predictions, with the possibility of many chemical structures matching to each formula. Here, newly developed quantitative structure-use relationship (QSUR) models are used to identify potential exposure sources for candidate structures. Previously, a suspect screening workflow was introduced and applied to house dust samples collected from the U.S. Department of Housing and Urban Development’s American Healthy Homes Survey (AHHS) [Rager, et al., Env. Int. 88 (2016)]. This workflow utilized the US EPA’s Distributed Structure-Searchable Toxicity (DSSTox) Database to link identified molecular features to molecular formulas, and ultimately chemical structures. Multiple QSUR models were applied to support the evaluation of candidate structures. These QSURs predict the likelihood of a chemical having a functional use commonly associated with consumer products having near-field use. For 3,228 structures identified as possible chemicals in AHHS house dust samples, we were able to obtain the required descriptors to apply the QSURs to 2,422 structures. The models returned at least one functional use prediction with a probability >80% for 1,421 chemicals (2,253 predictions). Given these new techniques, we are able to substantially reduce the number of potential structures to be analytically confirmed. For example, the molecular formula C18H34O2, had 33 possible structures based on our previous screening analysis. However, only 12 of those structures had predicted functional uses, 9 of which were associated with consumer products in US EPA’s Chemicals and Products Database (CPDat). These new models provide a critical filtering step for removing chemical structures not likely to be found in house dust, improving the interpretation of SSA findings and the prioritization of analytical resources.
Description:
Humans are potentially exposed to thousands of anthropogenic chemicals in commerce. Recent work has shown that the bulk of this exposure may occur in near-field indoor environments (e.g., home, school, work, etc.). Advances in suspect screening analyses (SSA) now allow an improved understanding of the chemicals present in these environments. However, due to the nature of suspect screening techniques, investigators are often left with chemical formula predictions, with the possibility of many chemical structures matching to each formula. Here, newly developed quantitative structure-use relationship (QSUR) models are used to identify potential exposure sources for candidate structures. Previously, a suspect screening workflow was introduced and applied to house dust samples collected from the U.S. Department of Housing and Urban Development’s American Healthy Homes Survey (AHHS) [Rager, et al., Env. Int. 88 (2016)]. This workflow utilized the US EPA’s Distributed Structure-Searchable Toxicity (DSSTox) Database to link identified molecular features to molecular formulas, and ultimately chemical structures. Multiple QSUR models were applied to support the evaluation of candidate structures. These QSURs predict the likelihood of a chemical having a functional use commonly associated with consumer products having near-field use. For 3,228 structures identified as possible chemicals in AHHS house dust samples, we were able to obtain the required descriptors to apply the QSURs to 2,422 structures. The models returned at least one functional use prediction with a probability >80% for 1,421 chemicals (2,253 predictions). Given these new techniques, we are able to substantially reduce the number of potential structures to be analytically confirmed. For example, the molecular formula C18H34O2, had 33 possible structures based on our previous screening analysis. However, only 12 of those structures had predicted functional uses, 9 of which were associated with consumer products in US EPA’s Chemicals and Products Database (CPDat). These new models provide a critical filtering step for removing chemical structures not likely to be found in house dust, improving the interpretation of SSA findings and the prioritization of analytical resources.
