[BUG] Gene.functional attribute not saved in SBML model

[wshao1]

Is there an existing issue for this?

• I have searched the existing issues

Problem description

When I knockout a gene, the gene.functional attribute is set to False. But when the model is saved/reloaded using write_sbml_model/read_sbml_model, the gene.functional attribute is reset to the default of True. I want the gene.functional attribute to be saved with each model.

A jerry-rigged solution is to save the attribute using _sbase_notes_dict in cobrapy/src/cobra/io/sbml.py:

_sbase_notes_dict(gp, {"functional": cobra_gene.functional}) --> after line 1344 in def _model_to_sbml

notes = _parse_notes_dict(gp)
if "functional" in notes:
    cobra_gene.functional = (notes["functional"] == "True")

--> after line 746 in def _sbml_to_model

I'm sure there's a better way to do this that aligns with SBML standards, so wondering if anyone who is familiar with SBML can fix this in a more elegant way.

Code sample

Code run:

# load textbook model and try KO'ing a gene
txtbook_model = load_model("textbook")
txtbook_model.genes.b0118.knock_out()
print(txtbook_model.genes.b0118.functional)

# save and reload textbook model
write_sbml_model(txtbook_model, 'txtbook_model.xml')
saved_txtbook_model = read_sbml_model('txtbook_model.xml')

# check if gene is still KO'ed (functional = False)
print(saved_txtbook_model.genes.b0118.functional)

Environment

### Package Information

Package	Version
cobra	0.29.1

Dependency Information

Package	Version
appdirs~	missing
black	missing
bumpversion	missing
depinfo~	missing
diskcache~	missing
future	1.0.0
httpx~	missing
importlib-resources	6.5.2
isort	missing
numpy	2.2.1
optlang~	missing
pandas	2.2.3
pydantic	2.10.4
python-libsbml~	missing
rich	13.9.4
ruamel.yaml~	missing
scipy	missing
swiglpk	5.0.12
tox	missing

Build Tools Information

Package	Version
pip	24.2
setuptools	75.1.0
wheel	0.44.0

Platform Information

Darwin	24.2.0-arm64
CPython	3.12.8

Anything else?

No response

[cdiener]

Agreed it would be cool to have this, but this is currently not supported by SBML AFAICT. The only allowed attributes for genes are the id and name essentially and nothing else. You can save functionality in the notes, but we would probably not do this here because the SBML standard is pretty strict about not storing parseable information in the notes. There were some discussions of new key-value annotations but I don't think this is available yet.

[wshao1]

OK thanks Christian! I'll stick with my workaround solution for now, and will keep an eye out for any SBML updates.

[Midnighter]

You might look at OMEX files which are essentially archives that include models and data files (like constraints or expression values).

[Midnighter]

Key values are in SBML FBC level 3, I think.

[wshao1]

Thanks for the suggestions Midnighter!

[cdiener]

The current FBC version implemented is still version 2 I think. Version 3 has not been released AFAICT and there seem to be no reference implementations.