MzTab Validation

[Lilferrit]

Automatically runs mzTab validation via github actions using a compiled jmzTab validator executable.

[Lilferrit]

I am not looking for a review just yet I just opened this PR to run the test jobs and test the jmzTab validator steps.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.03%. Comparing base (9e3b630) to head (e6f9185).

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #362      +/-   ##
==========================================
+ Coverage   91.48%   94.03%   +2.54%     
==========================================
  Files          12       12              
  Lines        1022     1022              
==========================================
+ Hits          935      961      +26     
+ Misses         87       61      -26     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[bittremieux]

Nice clean implementation. 👍 One minor style comment.

One more thing that you might investigate a bit more before merging though is whether the validation can still be extracted. I'm thinking something like a separate GitHub action, where it now says "Set up Java" and uses that specific action, that we might create a dedicated action that provides mzTab validation. I haven't done something like that myself yet, so I don't know how it would work or whether it'd be possible.

The immediate benefit is that the jar would not need to be included in the Casanovo repository, but rather in a separate repository that would implement this action. The bigger benefit, however, would be that we'd then be able to easily call this action for any tool that exports to mzTab, whereas now the validation is Casanovo-specific.

[bittremieux]

Hmm, question though: the change to export the precursor charge as an integer instead of a float in 9792584 doesn't seem to be retained in the final diff. Did that get reverted? And if so, the generated mzTabs should be non-compliant, so the test should fail. Can you double-check that the output is correct and the tests run successfully?

The precursor charge export fix should also be mentioned in the changelog.

[Lilferrit]

Hmm, question though: the change to export the precursor charge as an integer instead of a float in 9792584 doesn't seem to be retained in the final diff. Did that get reverted? And if so, the generated mzTabs should be non-compliant, so the test should fail. Can you double-check that the output is correct and the tests run successfully?

The precursor charge export fix should also be mentioned in the changelog.

I checked ms_io.py on the latest commit on this branch and the precursor charge fix is still there. At one point I removed the fix to make sure that the pipeline fails if the output file isn't compliant, but I simply put the fix back when I was done. This probably messed with the git diff. The changelog still has the fix mentioned in it as well.

[bittremieux]

Yes, you're right. Very weird, the diff doesn't even indicate that ms_io.py was changed, nor the changelog.