Combining intracellular models with a single model

[vincent-noel]

Hi all !

This is a PR on something I've been thinking about for some time: the possibility of having multiple types of intracellular models, and to have them interact. To achieve this, we need to modify a bit the implementation of Intracellular, mainly to change the variable in Phenotype, from an Intracellular pointer to a vector of Intracellular pointers.

This leads to a few more changes, such as having the pre/post intracellular update function belonging to the Intracellular object, and to add the Intracellular to a few method parameters (to be sure to have the "good" one).

Here I'm providing a quick example I build in the last days, which has an SBML ODE model for the TNF receptor binding and endocytosis and a Boolean model for the cell fate. To build it, use make spheroid_tnf.

For the ODE model, I'm using what I think is the latest implementation of roadrunner, based on #362 and #406. I had to modify a few things, such as introducing the Intracellular* parameter in many places, and I also changed the way the species values are modified before the update. So before thinking about merging this, we should focus on #362 and #406, and I can make a separate PR to have my changes too.

I am also working on example using the latest version of dFBA which is being developed at BSC by @marcorusc and @migp11, but I'll wait for them to submit their PR.

Let me know what you think !

[drbergman]

I think this is really cool! We should have a conversation around next updates to Intracellular together since, as you pointed out, this interacts in non-trivial ways with other open PRs. This feels like a use-case for a GitHub project...

[migp11]

This sounds great! Together with @marcorusc and @othmya, we are finishing the physicell-dfba paper, which includes several different use cases that can be
@vincent-noel have you reimplemented the TNF receptor model in SBML from this publication?

• https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2022.836794/full
  That is a very nice test case!
  Regarding the implementation of a hybrid model, including a Boolean model and an FBA-based metabolic model, I've been thinking and got some ideas based on @othmya last paper
• https://www.nature.com/articles/s41540-026-00669-4
  and another paper we did on AGS metabolism
• https://www.nature.com/articles/s41540-025-00586-y
  The nice thing about this potential use case is that there is already a Boolean model and a metabolic one.
  Open to discuss the topic :)
  Best

[vincent-noel]

@migp11 Indeed, this is your TNF receptor model. That was the simplest example I could think of, and it worked.

About the AGS model, that's a nice idea. I have some reading to do, and then let's talk about this.