5. ME-model Saving and Loading

There are currently 3 methods that can be used to save/load an ME-model using COBRAme: 1. As a full JSON file 2. As a reduced JSON file 3. As a pickle file

5.1. As a full JSON file

This is the recommended way to save, load and share COBRAme ME-models in full detail. This will include all of the model’s functionality and information (MEReaction, ProcessData, etc). It uses a defined JSONSCHEMA found in cobrame.io.

Saving and loading a full ME-model (me_model) as a JSON can be done using:

from cobrame.io.json import save_json_me_model, load_json_me_model
save_json_me_model(me_model, '[save_loc]/model.json')

Then loading can be done with

new_me_model = load_json_me_model('[save_loc]/model.json')

where new_me_model is of type cobrame.MEModel

5.2. As a reduced JSON file

Alternatively, ME-models can be saved as a COBRApy model. This storage type loses all the additional information contained in a full ME-model, but retains the stoichiometry of all the reactions. In other words, it behaves like an M-model with symbolic mu terms in metabolic coefficients and reaction bounds. Therefore it will give identical solutions compared to the full model, but all additional ME-model functionality will be lost.

Saving and loading a reduced Me-model (me_model) as a JSON can be done using:

from cobrame.io.json import save_reduced_json_me_model, load_reduced_json_me_model
save_reduced_json_me_model(me_model, '[save_loc]/model.json')

Then loading can be done with

new_me_model = load_reduced_json_me_model('[save_loc]/model.json')

where new_me_model is of type cobra.Model

5.3. As a pickle file

This is the quickest way to save a ME-model in full detail. It can be accomplished using python’s pickle dump/load methods. A ME-model named me_model can be saved follows.

import pickle
with open('[save_loc]/model.pickle', 'wb') as f:
    pickle.dump(me_model, f)

It can then be loaded with:

with open('[save_loc]/model.pickle', 'wb') as f:
    new_me_model = pickle.load(f)

This is not a recommended way to save a ME-model when sharing or for use over the long term as it can break when using different software versions.