GATK-DRAGEN standalone pipeline?Answered
I have been asked to try out your GATK-dragen pipeline. I found this and related pages:
And have been reading the docs to try and figure out if I can either:
1.) Get a list of the commands recommended to reproduce this pipeline on our system or
2.) Run the pipeline on my own system (ie. Not on Terra).
I have downloaded the .wdl from here:
but am total newbie when it comes to .wdl. Do you have any tips/tutorials/ideas about how I should go about testing the pipeline without having access to terra (or other cloud alternatives). For example, where should I source the containers and references that are needed?
This DRAGEN-GATK topic was just for the webinar in December, I am going to move your post into our Community Discussions -> General Discussion topic.
The resources for the DRAGEN-GATK pipeline are all in our gs://gcp-public-data--broad-references/ google bucket. Here is the README where we describe the resources for this pipeline: https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/README_dragen_gatk_resources.txt
Here is an overview of the main DRAGEN-GATK steps:
- Alignment with DRAGMAP using the masked reference file. https://github.com/Illumina/DRAGMAP
- In GATK, run CalibrateDragstrModel on your input bam. https://gatk.broadinstitute.org/hc/en-us/articles/4414602469531-CalibrateDragstrModel
- HaplotypeCaller per sample in DRAGEN mode using these arguments: --dragen-mode true --dragstr-params-path dragstr_model.txt
We don't directly support WDL on this forum A good place for getting started with WDL is looking through the docs at https://openwdl.org and https://github.com/openwdl/learn-wdl.
Please let me know if you have any further questions!
We have another helpful article I forgot about, which is a tutorial of the new DRAGEN-GATK steps: https://gatk.broadinstitute.org/hc/en-us/articles/4407897446939
I'm aware that germine line variant calling can be performed in Dragen mode now. Can Dragen-GATK also be used with Mutect2 for somatic variant calling (tumor-normal samples)?
It would be very helpful if you could link any webinars or tutorials which show how to run Dragen and Mutect2 on an HPC.
Hi Abhay Rastogi,
Thank you for posting your question to the forum! I'm sorry it took us so long to get back to you.
The DRAGEN-GATK options are only available in HaplotypeCaller. You can read more about this in another forum post here: https://gatk.broadinstitute.org/hc/en-us/community/posts/4411939432731-What-are-the-plans-for-expanding-DRAGEN-GATK-to-somatic-variants-
This was once a very useful resource for learning to work with sequencing data and implement quality scripts and pipelines.
The post highlights how annoying the site has become. You advertise a new product with fancy videos, images, and cloud-based pipelines but without linking to the description of how to use it. So, one has to write a post and ask for it. And while this post has thankfully been answered, nothing changed about the lack of a description. Best practices were once an awesome resource. Now, they are fragmented, incomplete, and don't show any code but refer to WDL pipelines because it is "easier" to read for people who have to work in bash anyway. It really, really is not more convenient to familiarize oneself with the ninth 'new and improved' descriptive language. Furthermore, links within pages link to the same page instead of a useful target; that's not annoying.
Hi Michael Olbrich,
The article linked above (https://gatk.broadinstitute.org/hc/en-us/articles/4407897446939) does provide an overview of the new steps in the DRAGEN-GATK pipeline, with example command lines for each tool.
The WDLs are the complete reference code for the pipeline, however -- they are what we run internally, and officially support externally. If you want to simply run the pipeline as we implemented it, they are your best option. And if you want a complete up-to-date reference for all the tools in the pipeline and their command-line arguments, they are also the best option. We've published many "quick-start" tutorials to help people learn to read/write WDLs, such as this one: https://support.terra.bio/hc/en-us/articles/360037117492-Overview-Getting-started-with-WDL
At the end of the day, a WDL is not much more than a series of tool command lines.
If you find any broken links in our documentation, please report them to us so that we can fix them.
Please sign in to leave a comment.