Conversation
Post release bump
Improve samplesheet validation
- correct error message for unexpected FASTQ count - require samplesheet info fields to be properly set - allow empty longitudinal_sample info field in samplesheet - fix check for duplicate samplesheet entries - improve samplesheet data parsing order - fix ESVEE version collection
Switch from BLAST to BWA-MEM for query sequence search
- BamTools to 1.5 - CIDER to 1.1 - COBALT to 2.2 - ESVEE to 1.2 - LILAC to 1.7.3 - LINX to 2.2 - ORANGE to 4.1.2 - PURPLE to 4.3 - REDUX to 1.2.2 - TEAL to 1.3.6 - VirusInterpreter to 1.7.2
Important! Template update for nf-core/tools v3.4.1
- note requirements for REDUX input discovery on object stores - promote compute resource to dedicated page - update information on RNA panel reference data
In order to run the 'test' profile on nf-core CICD, a large RunsOn instance is required with more 32 GB memory. We will eventually have many small tests but don't want these to also wastefully run on large instances. To avoid this situation, I split tests into different sizes where only the larger tests run on the up-sized instance.
Important! Template update for nf-core/tools v3.5.1
- adjustments to align with latest nf-core template release - update filetype and process links in documentation - update version references throughout
Apply Hartwig updates, fixes
Apply minor fixes and updates
Update tests and relevant GH Actions workflows / actions
Prepare release 2.3.0
|
Fixes issue with GH Actions where the Conda-base nf-test process expects and requires use of environment.yml otherwise a fatal error occurs in the currently support minimum version of Nextflow (25.04.0).
|
The failing GH Actions workflow for nf-test where Conda provides software is caused by:
The latter is intermittent and I found that it often wasn't an issue when attempting to replicate outside of the GH Actions runners. Given it is directly caused by the slow / unreliable connection of the server that hosts the Dfam database, this intermittent failure is difficult to address I was able to resolve the issue by hosting the Dfam database on a high-bandwidth serve, modifying the Conda package to point to this server, and then adjusting oncoanalyser to instead install the modified packaged (see this run). This also reduced environment creation time from ~40 minutes to <5 minutes As re-hosting the Dfam database on an unofficial server and using a non-standard Conda package is not hugely desirable, I am okay with proceeding where the nf-test ~ Conda failures in this PR are noted as being a problem with the upstream package / official servers |
Final release adjustments
ohofmann
left a comment
There was a problem hiding this comment.
LGTM, and looking forward to the new release!
- PAVE 1.8.2 - SAGE 4.2
Update SAGE and PAVE, software version collection
No description provided.