Analytical Choices Drive Toxicogenomic Potency Estimates: A Systematic Evaluation of Transcriptomic Points of Departure

Abstract

Omics technologies are increasingly integrated into next-generation risk assessment, yet quantitative toxicogenomics outcomes remain highly dependent on analytical choices, motivating a systematic evaluation of how bioinformatics workflows influence hazard characterization and transcriptomic Points of Departure (tPOD). Here, we applied five independent transcriptomics pipelines to a shared dataset of RPTEC-TERT1 kidney cells exposed to cisplatin across multiple concentrations and timepoints, comparing effects of pre-processing, benchmark concentration modeling, and pathway-based interpretation strategies. Across workflows, substantial variability was observed in gene-level benchmark concentrations (BMCs), primarily driven by differences in normalization, filtering, and especially the modeling software used. Despite this variability, convergence increased at later timepoints as transcriptional responses strengthened, with 24 h consistently identified as the most sensitive timepoint at the gene level. Aggregation of gene-level BMCs into pathway-based metrics reduced variability but did not eliminate it, with pathway definition emerging as a major determinant of sensitivity estimates. Notably, distinct pathway resources showed minimal gene overlap, and smaller, biologically coherent gene sets (e.g., co-expression modules and biomarker panels) produced lower and less dispersed BMCs compared with broader pathway annotations. Furthermore, direct modeling of pathway activity scores yielded systematically different sensitivity estimates relative to median-based aggregation, with method-dependent conservativeness influenced by pathway coverage and response strength. Overall, our findings demonstrate that both analytical workflow design and pathway selection critically shape toxicogenomic-derived potency estimates, highlighting the need for harmonized, transparent methodologies to enable robust application of transcriptomics in chemical safety assessment and regulatory decision-making.