Hi everyone, I'm Son, a 2nd year PhD student. I'm ...
# modules
s
Hi everyone, I'm Son, a 2nd year PhD student. I'm currently adding a new module: integron_finder. • When I do the PR, it passed all but the test with conda. • Then I checked again locally with nf-core modules test --profile conda, it resulted in:
Copy code
ERROR    nf-test failed due to differences in the snapshots
• I guess the problem comes from the differences between the conda and the container (even though I used the same version : bioconda::integron_finder=2.0.5 vs biocontainers/integron_finder:2.0.5--pyhdfd78af_0/https://depot.galaxyproject.org/singularity/integron_finder:2.0.5--pyhdfd78af_0, respectively). • Here is my PR: https://github.com/nf-core/modules/pull/8131 ? Then can I please ask for advices what should I do now ! Thank you so much in advance for your helps
i
This means that the output is different when Conda is used. Are you using the same version of the application in Conda as is included in the containers?
You can also link the PR if you have it
I see the link now, I'll have a look
s
Thank you so much for your answering @Igor Trujnara ! Yes, I've checked and apparently they are the same Sure here is the PR: https://github.com/nf-core/modules/pull/8131
i
Not even versions.yml matches; weird. Can you run the tool with Docker and Conda and then run diff on the output files? Maybe that will help pinpoint the issue
s
@Igor Trujnara I'm doing it, I'll update you when the results are available.
f
Maybe you can also output the yaml of the versions (as done in other modules) and then look at the conda output. I have the same issue with another module (and there it even only happens in the cI)
s
@Igor Trujnara Hi ! the diff results are here, as far as I can see here, the differences are between the version of dependencies used by either conda or container
@Famke Bäuerle Thank you so much for your suggestion ! I'm gonna check that.
i
Yes, looks like the issue is divergent dependencies. Can you try to package them with Seqera Containers to make sure that you get the same versions across both container types? It might require some manual Conda manipulation
s
@Igor Trujnara Thank you so much for your suggestion, let me try !