Closed mihai-sysbio closed 3 years ago
Very good idea.
Based on the previously implemented PR #173 in HumanGEM
, following checks may be included:
yamllint
cobrapy
importThere has been some progress on this. The suggestions from @Hao-Chalmers are already implemented as workflow with GitHub Actions in the automated-validation
branch, with results being automatically committed to that same branch in the results folder.
It looks like GitHub can schedule workflows only if they are available on the default branch (master), so it might be that the contents of the automated-validation
branch would have to be moved to a separate repo.
As mentioned in #16, the SBML fle should be tested with MEMOTE.
As mentioned in #13, the SBML should contain taxonomy name, taxonomy ID, and reference genome.
From the continued discussion in #16: perhaps one can devise a test to check the compatibility of compartment abbreviations as metabolite identifier suffixes
in cobrapy and cobratoolbox.
I opened a separate discussion on the topic of identifiers and semantic overloading at https://github.com/MetabolicAtlas/standard-GEM/issues/27.
For a clearer separation of concerns, the automated validation work has been promoted to a separate repository called standard-GEM-validation. The action points from this issue have been migrated to the new repository as distinct issues. Since all future work will be carried out in that repository, this issue will be closed.
This repository contains a branch (for)
automated-validation
. In short, the aim of that branch is to use GitHub Actions to validate all public repositories that followstandard-GEM
and output a machine-readable file with the results, eg which version of the GEM follows which standard,MEMOTE
results,cobrapy
testing. Documentation of the most up-to-date implementation is on the wiki. Ideas and feedback are welcome!edit: This could be implemented with eg Toil.