Item

Abstract

The identification of predictive patterns for toxicity from high‐throughput gene expression readouts in human in vitro assays is a fundamental goal of toxicogenomics. Reproducible and predictive assays based on stem cells have the potential to deliver such predictions in a more unbiased way than, for example, hepatocarcinoma cell lines. In this chapter, we describe the molecular characteristics of such a system built on a human embryonic stem‐cell‐derived assay for the liver. We report and extend published work that applied this assay for the first time to predict carcinogenicity induced by chemical compounds. We describe a bioinformatics approach that uses pathway‐wise expression patterns instead of gene‐wise expression patterns in order to identify predictive mechanisms for chemical carcinogenicity, and we show that such patterns are highly discriminative for the different toxicity classes of carcinogens.