Inter-laboratory study of human in vitro toxicogenomics-based tests as alternative methods for evaluating chemical carcinogenicity: a bioinformatics perspective.

摘要: The assessment of the carcinogenic potential chemicals with alternative, human-based in vitro systems has become a major goal toxicogenomics. central read-out these assays is transcriptome, and while many studies exist that explored gene expression responses such systems, reports on robustness reproducibility, when testing them independently different laboratories, are still uncommon. Furthermore, there limited knowledge about variability induced by data analysis protocols. We have conducted an inter-laboratory study for chemical carcinogenicity evaluating two human assays: hepatoma-derived cells hTERT-immortalized renal proximal tubule epithelial cells, representing liver kidney as target organs. Cellular were initially challenged thirty compounds, genome-wide was measured microarrays, hazard classifiers built from this training set. Subsequently, each system established three measurements using anonymized compounds. Data performed separate groups applying protocols reproducibility prediction hazard. As result, both workflows came to very similar conclusions respect (1) identification experimental outliers, (2) overall (3) re-classification unknown compounds respective toxicity classes. In summary, developed bioinformatics deliver accurate measures comparison studies, can be used guidance validation future order implement alternatives animal testing.