Hi everyone, does someone have experience with community modeling? I could use a pointer how to change compartments of all metabolites etc. (this seems to be kind of implicit if I understand correctly, so I tried changing the id to something with another ending (like '_XY' instead of '_c' )). So what I want to do is duplicate some metabolites and reactions into a new compartment, depending on which compartment they are in currently. So far, it only worked properly when I edited the sbml file in a text editor (which of course is an option...). Best, and thanks in advance! :-)
Need to change the name and Metabolite.compartment as well. Also see https://github.com/micom-dev/micom which will manage that for you (sorry for the shameless plug) 😃
Hey, I am trying to use four xml files to build a community using micom - no matter what I choose as value in the "id" column, I always get a ValueError: 'The model already contains a reaction with the id:', 'RXN_var1xy_1' - does micom somewhere use this "var" part as an identifier? Because that is one of the peculiarities of the reaction identifiers of my models.
Could you paste your taxonomy table?
There are the ones listed here: https://opencobra.github.io/cobrapy/tags/reconstruction/. There is also https://github.com/migp11/fastcore but I don't know if that is still maintained.
Hi I updated the bioconda recipe for cobrapy and it should work. Not official yet but would appreciate some testers. In a fresh environment you can install cobrapy with
conda install -c bioconda -c conda-forge cobra.
Preprint here: https://www.biorxiv.org/content/10.1101/637124v1
"Driven" is also a package from the DTU crowd that, if you're a power user, you could probably make work: https://github.com/opencobra/driven/tree/devel
Speaking of driven, @Midnighter are there any plans for further development of driven? Very few people seem to know it exists (name makes it difficult to find) but I'm sure many people would be willing to contribute.
Hello! I am working with Recon3D and HMR2 .mat files. I am loading them into cobrapy using cobra.io.load_matlab_model('recon3D.mat'). I can't figure out where the subsystem assignments are stored - any insights?
I have another model (humanGEM) in both .mat and .xml formats. The .xml file has subsystem information stored in "groups", whereas for the .mat file, the groups attribute is empty and I can't find subsystem information. Just wondering if my .mat files are missing the information or if I am loading the .mat files into cobrapy incorrectly...Thank you!
Hi @ablazier, the subsystem is an attribute directly on a reaction object (https://cobrapy.readthedocs.io/en/latest/autoapi/cobra/core/reaction/index.html#module-contents). SBML groups are unfortunatley, not yet fully supported.
Hi, this might seem a bit weird ... can someone click the link to lxml here: https://cobrapy.readthedocs.io/en/stable/io.html (3.1, SBML) and tell me where it takes them?
No idea what is up with lxml's site : |
I removed lxml as an optional dependency but as usual the docs are outdated :crying_cat_face: I'll try to get cobrapy into https://developers.google.com/season-of-docs/ next year.
Hey, I feel that we should refactor the summary code a bit. It took me quite a while to understand whats going on there. Some more comments and better variable names would go a long way. I do think that summaries should be immutable (see #900 for the what can go wrong otherwise). Also I always get
cobra/core/summary/summary.py:206 UserWarning: Setting float_format to anything other than None will cause nan to be present in the output. and really ugly output.
@cdiener I do agree that the present state of summary is a bit messy. @Midnighter 's #882 looked really clean and finishing that would be a nice change.
@cdiener I have another question concerning micom and types of community models - micom makes compartmentalized models, right? Does it automatically create carrier reactions between the compartments? If I do not get a flux solution, could it be due to efflux carriers missing? I get an 'infeasible' result at the moment (using cplex in a conda environment). Is there also a way to make mixed-bag models with micom? Thanks in advance ;-)
@mostueve_gitlab Yes every model has its own compartment and all of those are themselves embedded in a larger "medium" compartment that represents the environment. Micom will add an export/import reaction between the individual taxa and the medium for every external metabolite and an import/export into/from the medium as well. So of your metabolites have standardized names between all your models (true for Agora) they will be able to exchange them with each other. There is currently no support for mixed bag models mostly because it is unclear how you would integrate the individual abundances/biomasses in that case. If everything is equi-abundant you could use the join_models function in micom to get a mixed bag model...
If I want to build mixed bag models with cobrapy (my models all have the same biomass function), is the merge() function the way to go? Seems to work very well, but I am not sure if I understand all the implications in the docstring (such as left and right, constraints etc).
@mostueve_gitlab Yes, exactly cytosolic compounds would need export reactions for that.
Model.merge will duplicate reactions that are present in both models which might be problematic if what you really want is a pan-genome model. Micom's join_models uses the union of all reactions (based on IDs) which is a bit different from merge and sets the biomass as the average of all biomass functions. So it depends on your usecase and the particular biological assumptions you make.