Take tissue, for example

The hard science here is the docs know about layering, to the point of cell concentrations.  This is a fairly hard 'vertical', it translates into any other 3D system with little distortion. The equations on those color transitions are actually chemical counts. Not as accurate as Planck, but retty good.

So this coloring technique is already bound to the curvature of the equations, skin should close normally, and resolution down to the skin layer very good, as seen from three colored axis. The system, embedded in your home spread sheet, should have the resolution down to some small group, less than a thousand, at least as accurate as an old slide rule, and good enough for this technique.

So they can drill down to a thousand cells, get the likely concentrations of the four anti body types, and how they would be distributed in a unit sphere, a very big deal.  The relationship between the Blue, food, is defined for the system as deviations from white, it knows noting about how food is digested. But after the modeler makes a transition in food distribution via his external equations,. The constraint is any transition has to keep the deviations counts smoothly spread.  The modeler is sending amounts to the blue generator, to be parsed as according to deviations pending. So, after the ransition equation, the modeler returns food chunks of sub type a, fine, the model does not care how the equations are partitioning food.

Gang the models.
A three coloring of the skin does not prevent you from focusing on three coloring on protein interaction.  It is simply backing the coloring scheme farther and farther into the fundamental equations.  This is all, essentially, a computer geek trick, minimizing the transaction counts to stability when doing finite element.

One point to hammer home.
The modeler is modifying N density, the force to equalize that because N is a fuzzy constant belongs to the modeler. This is where the kinetic energy plays, the systems moves to nominal N density, to the substrate does. So the modeler, in an infection, is actually punching hole in the model, the model will step down appropriately, like a sandbox pit boss. Bankruptcy included but zero is never reached.