lower each compartment to it's own reactionsystem

[anandijain]

now that https://docs.sciml.ai/Catalyst/stable/tutorials/compositional_modeling/ compositional reactionsystems has matured, i think we may want to do this

it feels more natural and aligned with the sbml spec

[anandijain]

from @paulflang

the problem is that SBML reactions have no compartment attribute iirc. only species do. so you would have to infer the compartment of the reactions from the species. then you can group reaction networks by species compartment, if all participating species are in the same compartment. from that, you can create compartment-specific reaction networks. and the you can just smash in all the reactions where the participating species are of more than 1 different compartment into another reaction network. and than you can smash them all together with compose(), I guess. But the whole thing is probably easier said then done. Atm the highest priority for SBMLTk imo is to support bidirectional events. Cause this gives actually wrong results and ppl need that working.

I would hope that in SBML, the only reactions that have species in different compartments have the form A.a <-> B.a. and would be limited to 2 species max. so either all species of a reaction are in the same compartment, or it is a simple transport.

it could be very common to have some weird A.a + A.b -> B.c, but if i saw that i would rather just separate that into A.a + A.b <-> A.c and A.c <-> B.c.