A multi-study meta-analysis of the psoriasis transcriptome

Reproducible disease genes and pathways converging on the JAK–STAT / STAT3 axis. Meta-analysis of three whole-skin RNA-seq cohorts from recount3, lesional versus healthy skin, with random-effects pooling, pathway and TF-activity inference, cross-study co-expression, and single-cell validation.

3whole-skin RNA-seq cohorts
2,379meta-DEGs (FDR<0.05, |log2FC|>1)
264genes gained by pooling
+1.25STAT3 pooled log2FC (FDR 2.2e-7)
Part 1 · 42 slides

Findings

Cohort assembly, differential expression, random-effects meta-analysis, pathway enrichment and TF-activity inference, the STAT3 lead finding with its honest caveat, and target triage toward druggability. Speaker notes included.

Open →
Part 2 · 45 slides

Methods & verification

How the numbers were made and checked: recount3/Monorail provenance, the coverage-not-counts trap, QC metric derivation and exclusion rules, Scissor and deconvolution validation, the STAT3 α/β junction correction, and the reproduction contract.

Open →
Supporting document

Scissor whitepaper

Standalone write-up of the single-cell arm: cell states that track the psoriasis progression gradient, a portable Scissor solver, results, limitations, and next steps.

Open →

Scope. Research and informational analysis of public transcriptomic data (87 slides across two decks). It is not clinical guidance, and nothing here establishes a treatment effect or supports a care decision. Effect directions are reproducible across cohorts; magnitudes carry high between-study heterogeneity (median I² = 78%), which the slides report rather than smooth over. The STAT3 α/β isoform-switch hypothesis did not replicate and is presented as retired.

Static site · use / to move through a deck, g for the slide grid, n to toggle speaker notes.