broadinstitute / GATK-for-Microbes

BSD 3-Clause "New" or "Revised" License
22 stars 6 forks source link

Docker independent pipeline? #29

Open Guan06 opened 2 years ago

Guan06 commented 2 years ago

Hi,

many thanks for putting together the GATK-for-Microbes pipeline, seems very promising!

I am wondering is there a way to have a docker independent pipeline, since in our hpc, we cannot use docker.

Thanks and best, Rui

shebdon commented 1 year ago

I am also interested in the answer to this question.

shebdon commented 1 year ago

For those who want to run without (can't run with) docker/cromshell, here's what I've been able to do for the shortReads pipeline. It's likely applicable to other's as well.

The MicrobialGenomePipleline.wdl imports two other .wdl files also found in the repository. Apply the following changes to all of those files to get a non-docker version working.

  1. Remove the runtime { } blocks. These point to docker and are apparently optional.
  2. Replace all instances of /usr/gitc/ with an appropriate path for your setup.
  3. Redirect the import lines at the top of MicrobialGenomePipleline.wdl to your local, modified copies of the appropriate files.

With those changes made, instead of docker or cromshell you can download and use cromwell-XX.jar with java -jar /path/to/cromwell-xx.jar run /path/to/file.wdl -i /path/to/input.json

The input.json file provides "gatk_override" files which must be .jar files. With my setup the file ~/gatk/build/libs/gatk-package-4.3.0.0-36-gb7f24ff-SNAPSHOT-local.jar worked, but the path to gatk itself did not (I believe it's actually a python file not a java executable).

Hope this helps.