We reanalysed public RNA-seq data from the telencephalon of adult zebrafish after 14 days of chronic unpredictable stress. The deposited runs concatenate pooled libraries, hiding the replicates; we recovered the library boundaries from the block structure of unaligned reads and realigned to GRCz11 with splicing enabled. Unique alignment rose from 63.7–64.2% to 80.6–81.6%, and 275 genes were differentially expressed (FDR < 0.05), 101 of the 155 reported previously and 174 new, with effect sizes matching the published ones (ρ = 0.754). Against gene sets from an nr3c1 knockout, the data show a coordinated, mirror-symmetric shift in receptor-dependent transcription—the receptor-induced set down, the receptor-repressed set up (p ≈ 0.001 for both)—while the original study reported cortisol at control level; “receptor-dependent” denotes set membership, not receptor occupancy. Two artefacts constrain the reading of individual genes: epidermal content differs between groups while neuronal content does not, and the gene list is clustered in the genome far beyond chance. Removing the 85 flagged genes (of 275) leaves the receptor-level result intact (p = 0.0008 and 0.0014) but withdraws the conclusions on keratins and barrier structures. That removal addresses gene-level artefact, not sample composition, which six libraries cannot settle: a between-group difference in tissue composition remains an alternative explanation.
No takes yet. Share an insight, caveat, or question.
Shamshurin et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: