GitHub

Workflow's Outputs

The workflow generates a (potentially large) number of results files that can be a bit difficult to navigate at first.

In what follows, GROUP_NAME represents the value of a group if the groupby option was selected. If no groups are provided, the workflow sets it to all. Similarly PHENOTYPE is one of the provided phenotypes input.

GWAS Summary Statistics

The GWAS summary statistics are stored in files with names GROUP_NAME.PHENOTYPE.gwas.tsv. One for each group and phenotype.

Each output file stores variant-level information and has the following columns:

  • CHROM: The chromosome ID.
  • POS: The variant position.
  • ID: The variant ID.
  • ALLELE_0: The baseline allele. This should be the REF allele provided in the input PGEN files.
  • ALLELE_1: The effect allele, this should be the ALT allele provided in the input PGEN files.
  • ALLELE1FREQ: The ALLELE_1 frequency.
  • BETA: The effect size.
  • SE: The standard error.
  • LOG10P: The negative log10(p-value).
  • N: The number of samples used for this variant.

Furthermore:

  • SAIGE outputs the extra following: ALLELE1COUNT, MISSINGRATE, TSTAT, VAR.
  • REGENIE outputs theextra following: TEST, CHISQ, EXTRA.

Finemapping Summary Statistics

Similarly to GWAS results, finemapping summary statistics are stored in files with names GROUP_NAME.PHENOTYPE.finemapping.tsv. One for each group and phenotype.

The columns of those files are:

  • CHROM: The chromosome ID.
  • POS: The variant position.
  • ID: The variant ID.
  • REF: The reference allele from the input PGEN file
  • ALT: The alternate allele from the input PGEN file
  • LOCUS_ID: The GWAS lead variant ID identifying the locus.
  • PIP: The posterior inclusion probability.
  • CS: The credible set ID.
  • UNPHASED_R2: The R2 with the GWAS lead variant ID.
  • SUSIE_CONVERGED: Whether the SuSiE algorithm converged.

Plots

Finally, the workflow outputs a series of plots:

  • The GWAS plots, formatted as GROUP_NAME.PHENOTYPE.PLOT_TYPE.png. One for each group, phenotype and plot type (currently QQ, Manhattan).
  • Optional locus plots, formatted as GROUP_NAME.PHENOTYPE.LEAD_VARIANT_ID.locuszoom.png. One for each group, phenotype and locus passing the fine-mapping thresholds.

Meta Analysis

If the groupby and meta_analysis options are specified, the workflow will also produce meta-analysis outputs. These are similar to the above specified outputs with GROUP_NAME=META_ANALYSIS.

The GWAS output fill then contain the following additional columns:

  • DIRECTION: Summary of effect direction for each study, with one '+' or '-' per study
  • HET_ISQ: I^2 statistic which measures heterogeneity on scale of 0-100%
  • HET_CHISQ: Chi-squared statistic in simple test of heterogeneity
  • HET_DF: Degrees of freedom for heterogeneity statistic
  • LOG10P_HET: log10 of p-value for heterogeneity statistic

Results Location

All results are found in the output folder whose lcoation and organisation depend on the execution engine.

Local Execution with Cromwell

The outputs will be stored in the output folder defined by the final_workflow_outputs_dir variable specified in the cromwell's option file. The output folder is typically organised in subfolders corresponding to the workflow's tasks.

For example, for a grouped analysis with meta-analysis you should have the following subfolders:

  • call-gwas_group_plots: Containing plots for each analysed group.
  • call-gwas_meta_plots: Containing the plots for the meta-analysis.
  • call-merge_gwas_group_chr_results: Containing GWAS results for each group.
  • call-meta_analyse: Containing meta-analysed GWAS results.
  • call-merge_fp_group_chr_results: Containing finemapping results for each group.
  • call-merge_fp_meta_chr_results: Containing finemapping results from the meta-analysis stage using summary statistics.

UK Biobank RAP

In that case, the output folder is defined by the destination argument provided to dx run. Results will be found as a flat list in the output folder and all intermediate results are stored in the intermediate subfolder.