Steps to running the GWAS on ANL’s servers.
- Install Hail and its dependencies
- Filter genotype files
- Phenotype and Covariate Files
- Set up GWAS environment
Install Hail
This was done by Tom Brettin. Hail is working on the nucleus machine, but not washington at the moment.
Filter Genotype Files
Downstream analysis can be a lot faster with a smaller genetic dataset. We filtered for individuals in the brain imaging cohort, and for typical GWAS population conditions: white British ancestry, not related to others in cohort, good SNP call rate, etc.
The exact list of eids is at /vol/bmd/meliao/data/eids/intersection_brain_asthma.txt, which is so named because the sample filters wre taken from the ukbREST example asthma query. The filtering was run with plink using file filter_bgen_files.sh.
This filtering has mostly been done in plink2.
Phenotype and Covariate Files
Set up GWAS environment
There is a hail environment available on the ANL servers accessible by the command
conda activate /vol/bmd/software/condaenvs/hail