Convert Recon3DModel from L3V1 to L2V4

[Sai Teja]

Hi all,

         I would like to convert Recon3DModel from sbml L3V1 to L2V4. I've tried using the online SBFC converters, they return an error.

         Could someone please help me do the format conversion?

         I also tried out the sbml2sbml tool, but no success. 

      

         The L3V1 file can be accessed here - https://bit.ly/2FujzeK 

         Alternatively, Recon3DModel can be downloaded from https://www.vmh.life/#downloadview

Thanks,

Teja.

[Lucian Smith]

I am guessing that the error you are getting refers to the fact that this model has FBC information in it, which cannot be translated to L2v4.  If you are OK with that, you would need to translate it in a translator that has the option to strip out all package information first.  If that's indeed what you want to do, we can help with that, but I want to make sure that's indeed what you want first.  Basically, you'd get a model with a lot of reactions with no kinetic laws, no objective functions, and no FBC constraints.  It might still be useful for something, but it won't be an FBC model any more.

-Lucian

--
You received this message because you are subscribed to the Google Groups "sbml-interoperability" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbml-interoperab...@googlegroups.com.
To post to this group, send email to sbml-inter...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sbml-interoperability/e2d12b4e-8366-4c83-a8aa-2c9772a92e6e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Nicolas Rodriguez]

What is needed are the fbc converters that would create the SBML with COBRA like annotations. The only things that would be dropped are the genes annotation if there are some.

Nico

To view this discussion on the web visit https://groups.google.com/d/msgid/sbml-interoperability/CAHLmBr0UZVHpe8yc-NbkRnXth9-W%2BadNFGHdfNF%3Di7rd7rjgBA%40mail.gmail.com.

[Greg Medlock]

You could give the libsbml-based parser in COBRApy a try, which can write an SBML file without the FBC extension: https://github.com/opencobra/cobrapy/blob/devel/cobra/io/sbml.py

As far as I know, this would result in a L3V1 file without FBC information which another tool might then be able to convert to L2V4.

But like Lucian said, doing this is suboptimal. Are you working with another package that requires L2V4? If so, and the package is open source and/or still under development, I would make the developers aware that sticking to L2V4 is hurting usability (and will only do so more in the future).

Greg Medlock

To view this discussion on the web visit https://groups.google.com/d/msgid/sbml-interoperability/b6d028be-8165-44e7-39eb-e7a65ff86ade%40ebi.ac.uk.

[Sai Teja]

Hi all,

      Thanks for replying, 

      I need the resulting format in L2V4 needs to have the reaction names, the reaction itself, gene association rules, reaction bounds and the subsystem information, similar to the Recon2 L2V4 model at https://www.ebi.ac.uk/biomodels/MODEL1109130000#Files .

     The package I would be using is "MUFINS: multi-formalism interaction network simulator",  NPJ Syst Biol Appl. 2016. 

Thanks,

Teja.

[Matthias König]

Hi Sai,

I am one of the authors of the cobrapy SBML parsers.

Currently, the parser is writing SBML L3 with the fbc-v2 package.

The problem with SBML L2V4 is that there are a multitude of different variants on how to store bounds, objectives, objective_coefficients, ...

That is one of the main reason why the fbc package was developed (and it is highly discouraged to use SBML L2 for fbc, but rather use SBML-L3+FBC).

I am not sure what you want to use MUFINS for exactly. For me the project looks dead and there were no commits in the last 4 years

https://github.com/kierzek/MUFINS

There is one contributor which was probably a PhD student which now moved on, so even if you get the model in the right format, I doubt there will be any support with issues you will for sure run into.

The SBML parser code is here:

https://github.com/kierzek/MUFINS/blob/master/MUFINS1.0/pymet/src/Data/SBML.py

You could write your own parser or converter based on that.

For me, the main question is what analysis you want to perform with your constrainted-based model?

If it is something mainstream your chances are high that these methods exist in COBRA (matlab) or cobrapy (python).

What I can tell from the description of MUFINS "We extend the constraint-based modelling (CBM) framework by incorporation of linear inhibition/activation constraints" is basically extending the LP problem with linear constraints (which can be very easily done in cobrapy).

Many of the ~omics data integration routines of MUFINS are available in COBRA and cobrapy.

So it all breaks down to "What do you want to simulate? Which constraint-based algorithms do you want to use? What additional constraints do you want to add to the LP problem?"

For the linear activation/inhibition constraints you need the modifier information in your SBML, which is missing in RECON3D.

The omics methods like iMAT or GIMME are available in COBRA and to my knowledge also in cobrapy.

Best Matthias

[Sai Teja]

Hi Matthais,

   

     I would like to use the fastiMAT method which is implemented in MUFINS to get the activity state of each reaction, which solves LP problems instead of MILP per reaction of the model.

     I am using COBRA Toolbox already for getting the iMAT reconstructions. But for obtaining the reaction activity using fastiMAT is better I felt on Recon3, owing to the larger size of the model and faster compute time of fastiMAT. I would gladly use Cobra Toolbox for this along with the other analyses I’m doing, but I didn’t find any similar implementations there. 

Thanks for your suggestions,

Teja.

[Matthias König]

Hi Sai,

makes sense now.

Did you already run iMAT on Recon3D? It requires ~2*30000 MILP solutions which should probably run overnight for a single transcriptomics/proteomics dataset.

If you don't have too many omics datasets this should be feasible.

There are also other methods which are probably a bit faster than iMat, e.g. fastcore, but not in the range of fast iMAT (which only requires 1 MILP as I understand the algorithm).

You should have a look at

https://opencobra.github.io/cobratoolbox/stable/tutorials/tutorialExtractionTranscriptomic.html

and

A systematic evaluation of methods for tailoring genome-scale metabolic models

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5526624/

Be aware that reconstructions strongly depend on the used algorithm.

Best M

--
You received this message because you are subscribed to the Google Groups "sbml-interoperability" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbml-interoperab...@googlegroups.com.
To post to this group, send email to sbml-inter...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sbml-interoperability/0ba222da-e995-46f0-8b1c-bf676decffd5%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Matthias König, PhD.
Junior Group Leader LiSyM - Systems Medicine of the Liver
Humboldt Universität zu Berlin, Institute of Biology, Institute for Theoretical Biology
  https://livermetabolism.com
koni...@googlemail.com

https://twitter.com/konigmatt

https://github.com/matthiaskoenig
Tel: +49 30 2093 98435

[Sai Teja]

Hi Matthais,

    No, I couldn't run "full" iMAT on Recon3D using COBRA.

    I am able to use iMAT in the COBRA toolbox, but the implementation in COBRA solves only one MILP, gets one feasible solution, then generates a tissue-model and does not look for alternate solutions by solving repeated MILPs per reaction to call whether a reaction is included in the final iMAT model. The one MILP run takes few seconds for me, but I am unable to properly implement the MILP per reaction in COBRA as per the original iMAT paper, which takes significantly greater time (as you said, probably overnight).

Thanks,

Teja.

To unsubscribe from this group and stop receiving emails from it, send an email to sbml-interoperability+unsub...@googlegroups.com.

To post to this group, send email to sbml-inter...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sbml-interoperability/0ba222da-e995-46f0-8b1c-bf676decffd5%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.