Update to handle DRAGEN data

umccr / umccrise

:snake: DRAGEN Tumor/Normal workflow post-processing

https://umccr.github.io/umccrise/

MIT License

22 stars 8 forks source link

Update to handle DRAGEN data #74

Closed scwatts closed 3 years ago

scwatts commented 3 years ago

Changes:

Overhaul automatic input detection
Allow inputs to be provided via named arguments
Ignore timestamp for input BAMs and associated indexes
Enforce restriction of single bcbio batch or DRAGEN run
Prevent execution with bcbio data as input
Run BCFtools, SAMtools stats on non-bcbio input
Update DRAGEN reference QC files
Upgrade to NGS Utils 2.9.0 (includes updated DRAGEN support)
Upgrade to MultQC 1.12dev+umccrise-0.0.2 (includes additional umccrise patches)
Copy and rename QC files specifically for the MultiQC report to improve display
Implement custom sample ordering for MultiQC report

Please also see:

https://github.com/umccr/NGS_Utils/pull/1
New UMCCR MultiQC fork and changes

pdiakumis commented 3 years ago

Thanks so much for this Stephen! Looks good to me overall, just want to make sure we don't merge this with master, at least not yet, due to the current bcbio setup for the patient data. Would it make sense to have a parallel 'main' dragen branch?

scwatts commented 3 years ago

Thanks for looking over these changes

Yes, multiple branches is preferable here given that we'll likely be maintaining a bcbio release and a DRAGEN release for some time. Since we'll soon be running patient data on DRAGEN and ICA it seems to make sense to have commits from this PR on the default/base branch and to create a new branch for the bcbio release. I suspect that the bcbio release will be receiving only hotfixes going forward so having it on the non-default branch also aligns well with this strategy

One other detail to note is that changes in this PR do not strictly break compatibility with bcbio, though some updates will be required to run bcbio in a way that conforms with v1.2.4

scwatts commented 3 years ago

Force pushing to:

rebase on top of master
remove HISTORY.md changes while dealing with RCs
add new commits:
- replace use of 'tumour' with 'tumor'
- downgrade dvc/dvc-s3 to 1.11.6 (see extended commit message for details)
- bump MultiQC version to 1.12dev+umccrise.0.0.2
- remove stale documentation relating to QC file generation