search

MetabolicAtlas / standard-GEM

The standard for open-source GEMs on GitHub
https://www.biorxiv.org/content/10.1101/2023.03.21.512712
Creative Commons Attribution 4.0 International
18 stars 5 forks source link

name of folder with models #4

Closed edkerk closed 4 years ago

edkerk commented 4 years ago

Description of the issue:

What should be the name of the folder containing the model files?

Note that option 2 and 3 are in line with the style of the other folders (code, data).

Expected feature/value/output:

/models or /model

Current feature/value/output:

/modelFiles

BenjaSanchez commented 4 years ago

@edkerk I think model would be the best option, in line with e.g. memote repos but also more accurate, as the folder should contain just the one model but in different formats.

BTW have you guys seen https://github.com/opencobra/cookiecutter-memote ? Looks like it's more or less similar to the efforts being done here, maybe would be good to align both standards :)

haowang-bioinfo commented 4 years ago

@BenjaSanchez indeed the efforts better be aligned.

Those Configuration Values had been mostly implemented in HumanGEM as metadata, which probably could be defined explicitly in standard-GEM, aligning with cookiecutter.

edkerk commented 4 years ago

Seems only model gets +1 votes, and compatibility with the memote cookiecutter is a strong argument for going for model.

mihai-sysbio commented 4 years ago

@Hao-Chalmers there is indeed a need to have MIRIAM-style metadata at the model level. To me, neither implementation (config values, Human-GEM metadata) is perfect, but it sounds like there should be a general discussion around this in order to formalize something.
@BenjaSanchez I really liked what cookiecutter aimed to do - basically putting all the structure in place. And to that end, standard-GEM aims this too, while trying to increase transparency.

Summary: we go for model.