Initial concentrations not overwritten after preequilibration

[LeonardSchmiester]

What did you expect to happen?
I have a model, where I want to reset initial concentrations for some species (in my case different drugs) after preequilibration. In the SBML file, I have an initial assignment to the species, which is set to 0 for the preeq condition in the PEtab condition file and then to e.g. 10 for the simulation condition. I would have expected x0 in rdata for this species to be 10.

What has happened instead?
It seems like the initial concentration of the species is not overwritten with the value in the condition file. So in the above example x0 is 0. If I delete the preeq condition, x0 is set correctly.

To Reproduce
Here is a minimal example to reproduce the problem: minimal_example.tar.gz

AMICI version and system environment

• OS and version: Ubuntu
• AMICI interface: Python
• AMICI version: 0.11.8

[paulstapor]

Might be that this has something to do with model.setReinitializeFixedParameterInitialStates()? If this flag isn't set to True, model simulation will show exactly the behavior you described...

[LeonardSchmiester]

Thanks for the hint @paulstapor . This indeed fixed the problem for the minimal example. Will also try for the large model now.

[dweindl]

Okay, the problem here is that simulate_petab does not handle that correctly for initialAssignments, but only if you have a species name in the condition table.

[dweindl]

I am not completely sure how that should be handled (-> PEtab-dev/PEtab#481).

[LeonardSchmiester]

Okay, the problem here is that simulate_petab does not handle that correctly for initialAssignments, but only if you have a species name in the condition table.

Using the species Id indeed gives the results I expected.

[dweindl]

As this turns out to be slightly more complicated (PEtab-dev/PEtab#481), I'd suggest to only add support for simple initial assignments for now, i.e. those with math elements consisting of a single parameter that is not a target of other rules. The rest should trigger a NotImplementedError. Will fix this problem here. Anything beyond can be addressed elsewhere.