You are here:
Ontology-based Semantic Mapping of Chemical Toxicities
Citation:
Wang, R., C. Ives, AND S. Edwards. Ontology-based Semantic Mapping of Chemical Toxicities. TOXICOLOGY. Elsevier Science Ltd, New York, NY, 412:89-100, (2019). https://doi.org/10.1016/j.tox.2018.11.005
Impact/Purpose:
Under the new toxicology paradigm put forth by the NRC, there has been rapid progress in high throughput screening of chemicals and their omics characterizations in both lab and field settings. There is, however, a large gap between the molecular phenotypes generated from these efforts and apical endpoints of greater regulatory relevance, which are typically found at higher levels of biological organization. Comparative toxicity assessment across species is also an issue. Ontology-based semantic mapping was explored in this pilot study to evaluate if these issues could be addressed by leveraging both increasing amount of public phenomics data and previously published chemical exposure studies . The findings are very promising. As phenomics data grows in coverage and ontological annotation of toxicity responses becomes more automated, OS-Mapping will become a valuable data-mining tool to complement existing approaches in chemical toxicity assessment. This study should be of interest to the toxicology research community and OPPT/OCSPP POs.
Description:
This study was undertaken to evaluate the utilities and performance of ontology-based semantic mapping (OS-Mapping) in chemical toxicity assessment. Nineteen chemical-species phenotypic profiles (CSPPs) were constructed by ontologically annotating the toxicity responses from more than seven hundred published studies of ten chemicals (atrazine, bisphenol-A, cadmium chloride, chlorpyrifos, copper sulfate, cypermethrin, dioxin, 17α-ethynylestradiol, malathion, Tris 1,3-dichloroisopropyl phosphate) on six species (carp, Cyprinus carpio; fathead minnow, Pimephales promelas; mouse, Mus musculus; rat, Rattus norvegicus; trout, Onchorhynchus mykiss; zebrafish, Danio rerio). The CSPPs were semantically compared against over 29 thousand publicly-available phenotypic profiles of genes, KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways, and diseases based on a cross-species phenotype ontology. OS-Mapping could differentiate chemical toxicities among themselves, as well as within and across species. It also revealed cases of chemical by species interactions. Besides confirming the interspecifically similar MOAs (mechanisms of action) for a few same chemicals, OS-Mapping also generated novel insights into the MOAs underlying some seemingly different yet phenotypically highly similar classes of chemicals. The nature of a unified cross-species phenotype ontology and its representation of diverse knowledge domains allowed construction of a complete phenotypic continuum for EE2_FHM (17α-ethynylestradiol_fathead minnow) across biological levels of organization, which complemented a similar one derived from the Comparative Toxicogenomics Database but largely based on EE2-induced molecular phenotypes alone. Overall, we conclude that OS-Mapping offers a powerful approach to help bridge the gap between the molecular phenotypes of chemicals characterized by traditional omics and their apical endpoints of greater regulatory relevance, which are typically phenotypes found at higher levels of biological organization. OS-Mapping also enables comparative toxicity assessment among chemicals, both within and across species. Furthermore, semantic analysis of phenotypes can inform chemicals of their candidate MOAs when some of the chemicals of interest are less biologically characterized. A full phenotypic continuum based on OS-Mapping will also be conducive to future development of adverse outcome pathways. As phenomics data continues to grow in coverage and ontological annotation of literature becomes more automated, the power of OS-Mapping will be further enhanced.
