Dose-dependent differential gene expression provides critical information required for regulatory decision-making. The lower costs associated with RNA-Seq have made it the preferred technology for transcriptomic analysis. However, concordance between RNA-Seq and microarray analyses in dose response studies has not been adequately vetted.
Researchers at Michigan State University compared the hepatic transcriptome of C57BL/6 mice following gavage with TCDD using Illumina HiSeq RNA-Sequencing (RNA-Seq) and Agilent 4x44K microarrays using the same normalization and analysis approach. RNA-Seq and microarray analysis identified a total of 18,063 and 16,403 genes, respectively, that were expressed in the liver. RNA-Seq analysis for differentially expressed genes (DEGs) varied dramatically depending on the P1 (t) cut-off while microarray results varied more based on the fold change criteria, although responses strongly correlated. Verification by WaferGen SmartChip QRTPCR revealed that RNA-Seq had a false discovery rate of 24% compared to 54% for microarray analysis. Dose–response modeling of RNA-Seq and microarray data demonstrated similar point of departure (POD) and ED50 estimates for common DEGs.
There was a strong correspondence between RNA-Seq and Agilent array transcriptome profiling when using the same samples and analysis strategy. However, RNA-Seq provided superior quantitative data, identifying more genes and DEGs, as well as qualitative information regarding identity and annotation for dose response modeling in support of regulatory decision-making.