Encoding of the "or" when converting from SBMLqual to bnet

[nanls]

Hi,

I am using BioLQM to convert an SBMLqual file to bnet.

The input SBMLqual file describes a system of 3 nodes sA, sB, sX, and the only transition is sX = sA | sB. Here is the content of the math tag of the SBML describing this transition :

            <math xmlns="http://www.w3.org/1998/Math/MathML">
              <apply><or/>
                <apply><eq/>
                  <ci>sA</ci>
                  <cn type="integer">1</cn>
                </apply>
                <apply><eq/>
                  <ci>sB</ci>
                  <cn type="integer">1</cn>
                </apply>
              </apply>
            </math>

In the resulting bnet file, I obtained sX, !sB&sA | sB.

I know that (!B&A) | B is equivalent to A | B, but I am wondering why I don't obtain the exact equivalent of what is wrote in the SBMLqual. Is there any reason to have picked this form, or is it a weird bug ?

[aurelien-naldi]

Hi,

It's not a bug but a limitation of the implementation: bioLQM stores the logical rules as decision diagrams which are optimized for evaluating the rules but often differ from their original representation. It would be nicer for model conversion to keep track of the original function and to use it directly when possible, but it would require large changes to the data structures.

[pauleve]

To workaround this issue, you can use the colomoto.minibn Python module (part of colomoto_jupyter Python package), either to simplify a bnet file:

from colomoto import minibn
bn = minibn.BooleanNetwork.load("file.bnet")
bn = bn.simplify()
with open("file-simplified.bnet", "w") as fp:
     fp.write(bn.source())

or through biolqm:

import biolqm
m = biolqm.load("myfile.sbml")
bn = biolqm.to_minibn(m) # simplify by default
with open("file-simplified.bnet", "w") as fp:
     fp.write(bn.source())

[aurelien-naldi]

That's a good way to get nicer looking functions in the bnet output, but the functions may still be transformed compared to the original file (you may recover the same functions in some cases, but it is not guaranteed)

[nanls]

Thanks for these explanations and snippets. I will try! :)