Imported from yuw444/mCAs_WGS (
UKB/cisGWAS/AGENTS.md). Install upstream withnpx skills add yuw444/mCAs_WGS --skill cisGWAS. Copyright stays with the author.
cisGWAS - Cis-acting CN-LOH GWAS
OVERVIEW
Burden testing and common variant GWAS for cis-acting CN-LOH events. Largest subdirectory (34 files).
WHERE TO LOOK
| Task | Script | Notes |
|---|---|---|
| Extract gnomAD variants | run_extract_gnomAD_variants.sh |
VEP annotations, run on O2 cluster |
| Write burden masks | run_regenie_write_mask.sh |
Uses extractBgenVariants |
| Define burden masks | make_burden_masks.sh |
LoF + missense (PrimateAI-3D > 0.6) |
| Phase rare variants | run_phase_rare.sh |
SHAPEIT5 phase_rare_static |
| Annotate allelic shift | annotate_AS.sh |
Add AS tag from mCA calls |
| Burden GWAS | run_burden_GWAS.sh |
Fisher + Binomial tests |
| Common variant GWAS | run_common_var_GWAS.sh |
Batched meta-analysis |
| Read-based phasing | refine_phase.sh |
Phase variants relative to CN-LOH |
| FDR correction | apply_bh_FDR.sh |
Benjamini-Hochberg |
KEY TOOLS
parse_gnomAD_vep.cpp- Custom C++ VEP parser (only file in UKB with C++ source)extractBgenVariants- Fast protein-coding variant extraction from bgenphase_rare_static- SHAPEIT5 v5.1.1
WORKFLOW
extract gnomAD → write masks → phase rare → annotate AS → run GWAS → apply FDR
CONVENTIONS
- Uses PrimateAI-3D scores for missense annotation
- QC: Drop variants 10x more common in UKB vs gnomAD non_ukb_nfe
- LoF masks:
LoF.all_transcripts.missense8.0.001
