Answers are grounded in this paper and cite their sources.
Stress granule (SG) gene-set scores summarize transcription associated with stress-response machinery, but their stability across scoring definitions is uncertain. We compared fibroblast–epithelial expression contrasts for four fixed SG lists in two colorectal cancer single-cell cohorts containing 195,527 cells from 85 donors. Analyses used donor pairing, gene-wise standardized expression, within-cell ranks and raw-count pseudobulk summaries, with tissue and sequencing-depth sensitivities. At least 20 tumor cells per compartment were available for 15 and 29 paired donors. Pseudobulk SG-A contrasts were negative in both cohorts: median differences in mean log2 counts per million were -0.383 (95% confidence interval -0.493 to -0.319) and -0.493 (-0.614 to -0.384). Rank-based SG-B contrasts were also negative in both cohorts. In a post hoc comparison using identical depth-retained cells and weights, the second cohort had positive gene-Z estimates and negative rank-based estimates; the positive estimates did not meet the eight-test false-discovery threshold. Spatial comparisons depended on the fibroblast reference subtype, and the CosMx panel measured only 11 of 27 SG-B genes. RNA–protein correlations varied across genes, while organoid depletion calls were more frequent for selected translation-associated members than for the four core members. These findings establish reproducible relative-expression contrasts under specified definitions and show that their direction can change across scoring and depth conditions. Reporting the scoring definition, retained donor population, depth sensitivity and gene coverage together identifies the contrasts that reproduce and provides a defined basis for testing their relationship to SG formation.
No takes yet. Share an insight, caveat, or question.
Yang et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: