Lessons from Mining High-Throughput Transcriptomic Data Covering Multiple Cell Lines and Hundreds of Chemicals (ASCCT 2026)

Presentation by Logan Everett at the American Society of Cellular and Computational Toxicology 15th Annual Meeting on Oct 27th, 2026.High-throughput transcriptomics (HTTr) can measure the expression of all protein-coding genes across thousands of samples, allowing for broad evaluation in vitro of many mechanisms of action across multiple chemicals and concentrations. Concentration-responsive trends can then be aggregated—either within biologically-related gene sets or using statistical summaries of all gene responses—to yield transcriptomic points of departure (tPODs) that are robust and protective of in vivo hazards. However, the bioinformatic process for deriving tPODs from HTTr data involves multiple steps and a substantial number of user-specified parameters. There is a critical need to understand how these methodological choices impact overall results to inform best practices and ultimately support the adoption of HTTr in regulatory applications.Recent work has focused on the widely used software package BMDExpress v3 to evaluate the impact of different analysis parameters on tPOD estimation across several in vitro chemical screening studies. Starting from baseline parameters previously chosen for in vivo transcriptomic methods, ten distinct parameters were systematically altered to test their individual impact on the overall potency and robustness of the resulting tPODs. Additional comparisons were made to two alternate workflows used by federal research organizations that distribute public HTTr data. The results of these comparisons will be used to develop guidance for transcriptomic concentration-response modeling methods, thereby supporting harmonization of methods across decision contexts and increased confidence in the use of in vitro tPODs for chemical toxicity assessments.