Hello,
this is incomplete now but should bump several last points for compatibility with the FROG spec (we work in parallel on the cmbpy, cobratoolbox and cobrexa-based implementations).
The main points are:
I reverted the change that flux column should show the objective value -- it shows the FBA result, but the methodology needs to be changed.
I added/renamed a few fields into the metadata.json, mainly we agreed that everything should have an underscore and there should be a clear identification of the model filename (it may be different than location) and FROG version.
If there's no problem with that, I'd like to bump the version so that installing the result is easier.
What's open:
I still need to write the flux comparison (the check is basically: flux in one file is within tolerance of other file's feasible range, and vice versa). Seems like I'll finally need to learn numpy. :sweat_smile:
I'd love to add some fields (notably, we noticed that fixed amount of software fields may be insufficient already for some existing cases, so we added a whole sub-dict where people can get verbose-- see the current frog spec). Is there please a recommended way to start this? Notably, the field descriptions are all around the code with many copies (docs, schema, python class, ...); is any of that a master version that can be used to regenerate the rest? I started adding stuff into the class but then I realized that might be the wrong place, so better ask.
Thanks&best!
-mk
cc @rsmsheriff @ntung
(PS. on a side note, I finally managed to run this on our HPC; the main problem with the installation was basically broken versions of the HPC python-related stuff, plus one compat problem with dependence versions in cobrapy (will send a separate patch there))
Hello, this is incomplete now but should bump several last points for compatibility with the FROG spec (we work in parallel on the cmbpy, cobratoolbox and cobrexa-based implementations).
The main points are:
fluxcolumn should show the objective value -- it shows the FBA result, but the methodology needs to be changed.metadata.json, mainly we agreed that everything should have an underscore and there should be a clear identification of the model filename (it may be different than location) and FROG version.What's open:
softwarefields may be insufficient already for some existing cases, so we added a whole sub-dict where people can get verbose-- see the current frog spec). Is there please a recommended way to start this? Notably, the field descriptions are all around the code with many copies (docs, schema, python class, ...); is any of that a master version that can be used to regenerate the rest? I started adding stuff into the class but then I realized that might be the wrong place, so better ask.Thanks&best! -mk
cc @rsmsheriff @ntung
(PS. on a side note, I finally managed to run this on our HPC; the main problem with the installation was basically broken versions of the HPC python-related stuff, plus one compat problem with dependence versions in cobrapy (will send a separate patch there))