UKBiobank / DNANexus. Working with WDL and directories.

Question

[Pierre Lindenbaum]

Hi all,

I was recently given a paid access to a subset of ukbiobank that uses dnanexus to browse the data and run the WDL-based workflows. The documentation is huge https://documentation.dnanexus.com/ and, imho, is missing some real examples.

For now I've asked some questions to the support and in the community forum but it takes long to get an answer. So for a start , I want to setup a very simple WDL workflows that would

• STEP1 :collect all the VCF.gz files under a directory
  • STEP2: run a bcftools query in a region for each VCF
  • STEP3: concatenate all the outputs into one final FILE

so far my WDL workflow starts with:

version 2.0

workflow MY_WORKFLOW {
   input {
       Directory dir
(...)

with a classical file-based filesystem it would be easy to setup a json file describing the parameters

(...)
"stage-common.bed_file":"/path/to/my.bed",
(...)

But for dnanexus it must be a json object like:

"stage-common.bed_file":{
   "$dnanexus_link":{
      "project":"project-12345",
      "id":"file-12345"
      }
   },
(...)

But how should I specify a Directory ? I looks like the directories don't have any ID ?

Thanks !

( cross posted: https://community.dnanexus.com/s/question/0D5t000004RZO4iCAH/wdl-how-to-specify-a-directory-in-the-json-input )

[Pierre Lindenbaum]

OK, I think I found the solution: https://github.com/lindgrengroup/ukbb-meta/blob/8bb9ed69f61866563bb10f7ee833b588ab45ce5f/wdl/tasks/get_link.wdl