TWAS of atopic dermatitis across 8 GWAS cohorts — S-PrediXcan/S-MultiXcan outputs and harmonized summary statistics

Transcriptome-wide association study (TWAS) results for atopic dermatitis (AD) across eight public GWAS cohorts covering European, East Asian, and Aboriginal Australian populations. The deposit contains the novel outputs produced by the pipeline at github.com/JerryIshihara/atopic-dermatitis-genetic-variant-study plus the harmonized summary statistics (hg38) used as input to the imputation step, so downstream users can run S-PrediXcan / S-MultiXcan or meta-analyses without repeating the 1-hour imputation pass. Cohorts (see manifest.csv): bbj-a-90 (Ishigaki 2019, East Asian, N=212,036), ebi-a-GCST90018564 (Sakaue 2021, East Asian, N=168,103), ebi-a-GCST90018784 (Sakaue 2021, European, N=481,299), ebi-a-GCST90027161 (Sliz 2021, European, N=796,661), finn-b-L12_ATOPIC / ATOPIC_STRICT / ATOPIC_STRICT_REIMB (FinnGen 2021, European), ieu-b-5145 (Budu-Aggrey 2023, Aboriginal Australian, N=864,982). Pipeline: OpenGWAS VCF → standardized table → harmonize (hg19→hg38 liftover) → summary imputation against 1000G EUR LD panel (22 chr × 10 sub-batches = 220 jobs/trait) → merge → S-PrediXcan across 49 GTEx v8 mashr tissues → S-MultiXcan cross-tissue. Based on MetaXcan (Barbeira et al. 2018) and summary-gwas-imputation (Barbeira et al. 2019). Files: 1 manifest, 8 standardized GWAS tables, 8 harmonized GWAS, 8 merged imputed summary stats, 8 S-MultiXcan gene-level results, 8 × 49 = 392 S-PrediXcan per-tissue results. ~7.3 GB total. Not included (already on Zenodo under their own DOIs): GTEx v8 mashr prediction models (10.5281/zenodo.3518299), 1000 Genomes LD reference panel (10.5281/zenodo.3657902). Reproducing these results: clone the repo and run bash gwas-twas-analysis/scripts/twas/run_all_ad.sh. The imputation step (Step 2) is long; a GCP-autonomous deployment helper is included.