Conversation
… same as BAM/BAM_REDUX/BAI files
…lude run processes excluded by default
… resource creation workflows
Co-authored-by: Stephen Watts <scwatts@users.noreply.github.com>
…y from workflow.manifest.version
…ionableCodingPanel and CoverageCodingPanel files
Add reports to tower.yml
Force install of compatible icu version 73.2, the newer 75.1 version available in conda-forge can be installed with older the r-stringr and r-stringi recipes but is incompatible.
…or-fastq-data Support relative paths for fastq data
All WiGiTS tools that accept the log_level argument support the 'OFF' value except CHORD, which causes oncoanalyser to crash. Therefore, this log_level / hmftools_log_level value is not allowed.
Apply minor fixes and updates
Prepare release 2.2.0
|
There was a problem hiding this comment.
The pipeline looks well written and organised. I read through the files but I haven't spend too much time since I am not familiar with most of the tools implemented here.
I am testing the pipeline and I will let you know if I encounter major errors/issues.
I have just few comments:
- As I understood you only have one nf-test for stub mode. shouldn't be good to include more?
- You have many modules in local. it would be good to consider to move some of them in nf-core/modules since other developers might find it useful.
- nf-core AWS fill size tests fail. it needs to be fixed
great job!
The GH Action self-hosted RunsOn instances do not have sufficient disk space to execute non-stub tests (currently 80 GB of EBS volume according to these docs: https://runs-on.com/configuration/job-labels/#disk). Also using nf-test tags as providing paths to desired tests is not honoured when --changed-since is supplied.
Reviewer recommendations and other adjustments
|
Thank you for the review @danilodileo! I've now added the We're happy to explore making these tools more easily accessible to the community, though in their current state they are strongly interdependent on one another and running them in a standalone way isn't really possible, so the modules aren't individually that useful to others I confirmed the AWS full sized test is also failing (though only on manual trigger), and this appears to be caused by some unspecified error during submission of the Seqera Platform job. Looks like it's a known issue with the current advice to ask someone from the nf-core team to trigger the job once the release is completed |
|
The newly failing GH Action workflow that runs I have manually tested the To confirm this wasn't related to oncoanalyser itself, I also built a minimal reproducible example. See here for more information: Since the |
ohofmann
left a comment
There was a problem hiding this comment.
I am okay with the proposed changes — thanks for all the work (and the external reviewers).
No description provided.