You are here:
Pathway-Based Concentration Response Profiles from Toxicogenomics Data
Citation:
Elloumi, F., R. JUDSON, D. J. DIX, I. A. SHAH, T. B. KNUDSEN, D. REIF, A. V. Singh, L. Zhen, AND F. Wright. Pathway-Based Concentration Response Profiles from Toxicogenomics Data. Presented at Society of Toxicology Annual Meeting, Baltimore, MD, March 15 - 19, 2009.
Impact/Purpose:
We applied this approach to data from a concentration-response study of rat primary hepatocytes treated with Phenobarbital, RU-486, or WY-14643, which are inducers of xenobiotic metabolism pathways via nuclear receptors. Cells were exposed to 4 different concentrations of each chemical, with 5 biological replicates per concentration-chemical combination. Although a total of 3^(N-concentrations) unique signatures was possible, the observed number of pathways and concentration-response signatures observed with high reproducibility was much lower. Affected pathways included expected results on xenobiotic metabolism, and the signatures gave a direct reading of the LOC at which gene expression in these pathways was significantly perturbed.
Description:
Microarray analysis of gene expression of in vitro systems could be a powerful tool for assessing chemical hazard. Differentially expressed genes specific to cells, chemicals, and concentrations can be organized into molecular pathways that inform mode of action. An important parameter in this hazard assessment is the lowest observed concentration (LOC) at which a particular pathway is perturbed. We present a method to derive LOC using microarray-based pathway analysis, and assess the reproducibility of results across biological replicates. Concentration response of pathways was modeled by clustering genes across concentrations, and then calculating enrichment across pathways. Because clustering continuous gene expression data can make comparisons between replicates difficult, we discretized fold-change to one of 3 values: 0 (no change), + (increase), or - (decrease). Genes with the same ordered sequence or “signature” of discretized values across concentrations were grouped together for pathway enrichment analysis.