feat: python I/O functions + BiGG compliance flag

[BenjaSanchez]

Main improvements in this PR:

• Features python functions for loading and saving the model (read_yeast_model and write_yeast_model) without needing to specify the model path. Usage, as indicated now in the README file, is as follows:

  import ComplementaryScripts.io as io
  model = io.read_yeast_model() # loading
  io.write_yeast_model(model)   # saving

  For this to work, a .env is required in the repo's root (as also indicated in the README file).
• Features a requirements.txt file for ensuring a reproducible environment when performing simulations in Python.
• To fully address compatibility of identifiers with BIGG database #172, read_yeast_model features a make_bigg_compliant flag for loading the model with BiGG ids instead of regular ids. If chosen, the BiGG annotation in the model + the BiGG dictionaries in ./ComplementaryData/databases are used for the conversion. Some additional info:
  • ids will have a _copyN suffix if they are repeated in the model (as ids should be unique).
  • ids not found will be kept the same as in the model (but in the case of metabolites, converted to the met_comp standard).
  • Original ids are moved to notes.
  • Compartments are made BiGG-compliant with the dict {"er":"r", "erm":"rm", "p":"x"}
  • Compartment info is removed from metabolite names.

NOTE that in each commit I saved the model with write_yeast_model to confirm changes were as expected, but in the last commit I reverted the model to be as the original one in devel, so no model changes are performed here.

NOTE ALSO that although the model can be saved in cobrapy format keeping all fields, it changes considerably the structure of the .xml file and hence should not be done if contributing to this repo until issue #218 is solved.

I hereby confirm that I have:

• Tested my code with all requirements for running the model
• Selected devel as a target branch (top left drop-down menu)

[hongzhonglu]

Hi @BenjaSanchez I am not sure whether errors exist when load model using python, i have used the followed self-function to correct some errors:
def correctSomeWrongFormat(model0):
"""
This function is used to correct some wrong format when read yeastGEM model from cobratoolbox
"""
for met in model0.metabolites:
met.id = met.id.replace('91', '_')
met.id = met.id.replace('93', '')
#for reaction in model0.reactions:
for gene in model0.genes:
gene.id = gene.id.replace('45', '-')

return model0

[edkerk]

I have no experience with running cobrapy, so I'm unable to review this functionality.

[BenjaSanchez]

@hongzhonglu the changes you are performing are not needed anymore with the latest versions of cobrapy, as they convert the non-ASCII characters by default. Are you testing the PR with the proper environment? please update your cobrapy version, or even better, pip install -r requirements.txt in a fresh environment to make sure we have the same setup :)

[hongzhonglu]

Hi @BenjaSanchez, Thanks for your information. I will test it.

[BenjaSanchez]

@hongzhonglu is the setup working now for you? :)

[hongzhonglu]

hi Ben, could you replace this package named "dotenv" as i found it is can't be installed by pycharm?

[BenjaSanchez]

@hongzhonglu the package is called python-dotenv, as you can see in requirements.txt (no package called dotenv). Pycharm should not have any problem with python-dotenv.

[hongzhonglu]

Hi Ben, thanks for your remind!
I meet new issue for the two steps:
dotenv_path = find_dotenv()
REPO_PATH = dirname(dotenv_path)

when run the above code, REPO_PATH is "", thus the followed commond can not be run. Why not directly set the whole path of model as the input for function "read_yeast_model"?

[BenjaSanchez]

@hongzhonglu did you follow the new installation instructions as the README now states? You should create an empty .env file in the repo root:

type nul > .env                     # creates a .env file for locating the root

The idea of using dotenv is precisely to avoid specifying the path depending of where in the repo we are, as that (absolute) path never changes. This makes it easier & safer for the user to read/write, as it requires less typing and avoids mistaking the path and saving the model somewhere undesired.

[hongzhonglu]

Now, i can run it!