Some SBML formats (level 3?) are not read correctly about cobrexa.jl

[exaexa]

This SBML uses a parameter-dependent definition of flux upper and lower bounds:

    <listOfParameters>
      <parameter id="FB1N1000" value="-1000" constant="true"/>
      <parameter id="FB2N0" value="0" constant="true"/>
      <parameter id="FB3N1000" value="1000" constant="true"/>
    </listOfParameters>

later:

      <reaction metaid="R_26DAPLLAT" id="R_26DAPLLAT" name="L,L-diaminopimelate aminotransferase"
       reversible="true" fast="false"
       fbc:lowerFluxBound="FB1N1000"
       fbc:upperFluxBound="FB3N1000">

This is not super-convenient because there are then multiple sources of the lower/upper bounds. I'll see what I can do.

[htpusa]

This looks to be the format output by the COBRA Toolbox function writeSBML.

[exaexa]

Do we need to represent the information about the parameter names, i.e. know that the value "-1000" is in fact named FB1N1000 ?

[htpusa]

I don't think so. I don't even know what "FB1N" is supposed to mean.

[exaexa]

OK, hopefully I'll be able to find the right SBML functions to at least get the FB1N :D

[exaexa]

Hm, do the FBC constraints have any units? It doesn't seem so to me, but that's very different from the careful handling of units that can be seen with non-FBC specified fluxes...

[htpusa]

It might help to look at the COBRA Toolbox functions writeSBML and readSBML (and readCbModel). COBRA outputs this format by default.

[htpusa]

They should have units? But I agree that the file doesn't really say that. Is it implicit?

http://co.mbine.org/specifications/sbml.level-3.version-1.fbc.version-2.release-1.pdf

[exaexa]

that's a bit opaque to me, for example your model has a unit definition in it (mmol per ...), but the unit isn't actually used anywhere. (well, technically the units are completely irrelevant for this kind of stuff, so whatever).

Can you try this commit? LCSB-BioCore/SBML.jl@86df302

[htpusa]

Bounds look to be working, objective is still empty.

[exaexa]

fixed now, I'll make a PR with that

[exaexa]

Closed by LCSB-BioCore/SBML.jl#23