saige-pipelines

Running GWAS

How to run SAIGE GWAS with Cromwell
This in an example scenario creating new phenotypes in R7 and running those

1. Create a covariate/phenotype file that contains your phenotypes. E.g. get gs://r7_data/pheno/R7_COV_PHENO_V2.txt.gz, add phenotypes to that (cases 1, controls 0, everyone else NA), and upload the new file to a bucket
2. Create a text file with your new phenotypes one per line, e.g.
   my_phenos.txt

   PHENO1
   PHENO2
   

   and upload the file to a bucket.
3. Clone this repo git clone https://github.com/FINNGEN/saige-pipelines
4. Cromwell requires subworkflows be zipped: cd saige-pipelines/wdl/gwas/ && zip saige_sub saige_sub.wdl saige_summary.wdl
5. Change saige.null.phenofile in saige.json to the file from step 1
6. Change saige.phenolistfile in saige.json to the file from step 2
   6.1. Use "saige.traitType": "binary" or "saige.traitType": "quantitative" depending on whether your traits are case/control or continuous
   6.2. Use "saige.analysisType": "additive" or "saige.analysisType": "recessive", "saige.analysisType": "dominant" or "saige.analysisType": "het" - additive being regular GWAS.
7. Connect to Cromwell server
   gcloud compute ssh cromwell-fg-1 --project finngen-refinery-dev --zone europe-west1-b -- -fN -L localhost:5000:localhost:80
8. Submit workflow
   8.1. Using the web interface
   8.1.1 Go to http://0.0.0.0:5000 with your browser
   8.1.2 Click /api/workflows/{version}
   8.1.3 Choose wdl/gwas/saige.wdl as workflowSource
   8.1.4 Choose the edited wdl/gwas/saige.json as workflowInputs
   8.1.5 Choose wdl/gwas/saige_sub.zip as workflowDependencies
   8.1.6 Execute
   8.2. Or with https://github.com/FINNGEN/CromwellInteract
9. Use the given workflow id to look at timing diagram or to get metadata
   http://0.0.0.0:5000/api/workflows/v1/WORKFLOW_ID/timing http://0.0.0.0:5000/api/workflows/v1/WORKFLOW_ID/metadata
10. Logs and results go under
    gs://fg-cromwell/saige/WORKFLOW_ID, plots gs://fg-cromwell/saige/WORKFLOW_ID/call-test_combine/shard-*/**/*.png, summary stats and tabix indexes gs://fg-cromwell/saige/WORKFLOW_ID/call-test_combine/shard-*/**/*.gz*

Docker file creation for R6 GWAS

Same image used for R7 GWAS

git clone https://github.com/FINNGEN/saige-pipelines
cd saige-pipelines
git clone https://github.com/weizhouUMICH/SAIGE -b finngen_r6_jk
docker build -t gcr.io/finngen-refinery-dev/saige:0.39.1-TAG -f docker/Dockerfile_SAIGE_GWAS .

Conditional analysis for genomewide significant regions.

wdl/saige_conditional_full.wdl and corresponding .json scan for genomewide significant regions and then performs conditional analysis on those regions, adding significant variants as covariate and iterating on that until no significant variants are left.

If you want to run conditional analysis without scanning for gw-sig loci from results files, you can use saige_conditional.wdl/.json directly. It needs configuration file which can be greated using scripts/generate_conditional_analysis_config.py. See scripts/generate_conditional_analysis_config_examples.sh for example commands.

Output files

PHENOTYPE.REGION.independent.snps files

Summary of top snp conditional statistics after conditioning

Columns:

• SNPID variant
• BETA original beta
• SE original se
• p.value original se
• BETA_cond beta after conditioning
• SE_cond se after conditioning
• p.value_cond p-value after conditioning
• Conditioned_on variants used in conditioning

PHENOTYPE.REGION_n files

Summary statistics of all snps in the region after conditioning on n snps.

The condition snp corresponds to the line in .independent.snps file

Columns:

• CHR
• POS
• rsid
• SNPID
• Allele1
• Allele2
• AC_Allele2
• AF_Allele2
• imputationInfo
• N
• BETA
• SE
• Tstat
• p.value original p-value using SPA estimator (you want this p-value)
• p.value.NA original p-value using normal approximation (don’t use!)
• Is.SPA.converge did the model converge
• varT original t statistic
• varTstat original variance of t-statistic
• Tstat_cond t-statistic after conditioning
• p.value_cond p-value after conditioning
• varT_cond t statistic variance after conditioning
• BETA_cond beta after conditioning
• SE_cond standard error after conditioning