accessing list of entities in a sample set in WDL

• 

  Jason Cerrato

  • December 06, 2021 21:39

  Hi Stephanie,

  Thanks for writing in. Samples in a sample_set are typically interpreted as an Array[File] in a WDL. I believe it should be possible to write a task that takes the Array[File] input and writes the contents to a list of type File. That File would be output for that task, and input for the task you want to feed it into.

  Alternatively, I wonder if you can pass that Array[File] into your command directly. What do you think?

  Kind regards,
  Jason

  0
• 

  Jason Cerrato

  • December 13, 2021 16:15

  Hi Stephanie,

  We haven't heard from you in a while so we just wanted to check-in. Did any of the suggestions from my previous message help? Do you have any questions I can help answer at this time?

  Kind regards,

  Jason

  0
• 

  Stephanie Gogarten

  • December 20, 2021 22:20

  I tried the simplest scenario I could think of, where the input samples are an Array[File] and the file list is written to an output file with `cat`:

  https://github.com/UW-GAC/primed-bcftools/blob/main/bcftools_stats.wdl

  However, when I try to run this workflow on Terra, it immediately fails, with no error message or any indication of what is going wrong. I feel totally lost when trying to write a WDL that works with data tables. Is there any documentation or examples that address how to use data tables in a WDL? I've found documentation on how to use data tables in an existing workflow, and documentation on how to write a WDL, but I haven't been able to find anything that combines the two.

  0
• 

  Stephanie Gogarten

  • Edited December 21, 2021 01:24

  I figured this out, and my confusion with the above answer came from the fact that I was trying to identify the "sample_id", not a sample-specific filename (which does not even exist in my use case, as I have multi-sample files). "sample_id" can be referenced in the WDL in an Array[String]:

  input {

  Array[String] sample_list

  }

  Then in the WDL task, I can access the contents of this String Array and write it to a file:

  command {

  echo ${sep=' ' sample_list} > file.txt

  }

  When I run the workflow, the input sample_list is specified as "this.samples.sample_id" where the root entity type is a sample set.

  0

Please sign in to leave a comment.