Citation:

Karmaus, A., P. Kothiya, S. Watford, R. Thomas, AND M. Martin. Customizing the Connectivity Map Approach for Functional Evaluation in Toxicogenomics Studies (SOT). Presented at SOT National Meeting, Baltimore, MD, March 13 - 16, 2017. https://doi.org/10.23645/epacomptox.5176657

Impact/Purpose:

Poster presentation at SOT in Baltimore, MD

Description:

Evaluating effects on the transcriptome can provide insight on putative chemical-specific mechanisms of action (MOAs). With whole genome transcriptomics technologies becoming more amenable to high-throughput screening, libraries of chemicals can be evaluated in vitro to produce large toxicogenomics datasets. However, developing a systematic approach for linking transcriptional changes to MOA has been challenging. This study presents a connectivity map (CMAP) inspired methodology to conduct gene set enrichment analysis using toxicogenomics datasets to identify putative MOAs for chemical-mediated effects. Our CMAP approach utilizes a reference database of differential expression profiles and a rank-based permutation test for identifying enriched molecular targets. This approach requires that profiles in the reference database represent a diversity of perturbations that are mapped or annotated to molecular targets. To satisfy these requirements, we established a reference database of ~900 whole-genome expression profiles from the original CMAP effort (Lamb et al., 2006) and annotated the chemical perturbagens to 86 unique targets. To evaluate the new custom CMAP approach, 34 chemicals were selected that encompass multiple MOAs including nuclear receptor agonists/antagonists, enzyme inhibitors, and chemicals interfering with cell integrity (tubulin disruption). MCF7 and HepaRG cells were treated with three concentrations of each chemical for six hours, and changes in whole genome expression were quantified using Affymetrix microarrays (De Abrew et al., 2016). Z-score distributions were used to identify differential gene expression (using a cutoff of z-score > 2) and profiles were matched using JG scoring (Jiang and Gentleman, 2007). Finally, a rankbased permutation was applied to identify targets enriched among the significantly associated reference profiles. Of the 34 chemicals evaluated, 17 had MOAs that were not sufficiently represented in the reference database, 11 were correctly and significantly matched to their putative target, and six targets/mechanisms of action were not correctly identified. By integrating molecular target annotation and rank-based permutations for targets, this adapted CMAP approach can help identify putative MOAs for chemical-mediated effects using toxicogenomic data. This abstract does not reflect EPA policy.

URLs/Downloads:

Record Details: