Getting different answers between Graphic and Textual versions of the same model

Phoenix Certara Forums Modelling and Simulation
1

Hi,

I need to start leveraging the PML language to properly setup my model. I write up my equations with concentration. However, when WinNonlin converts my model the equations use amounts. I did a small test comparing the abundance formulation with my equations noticed when i convert my graphical model to a text version.

I am getting very different answers between the too models. Now the difference between the rate constants can be attributed to a difference in units. The volume should be the same though and this is what has me alarmed.

I am not sure handles the number for the error term (graphics vs text ), but that is really the only thing I can think of that could drive the difference.

The reason why I want to write it in this fashion is I am looking to do something similar to the PML school lesson on metabolites. They use MM kinetics which uses concentration. This should be ok but I am a bit concerned about the difference answers using the two modes.

I have attached the workbook.

Any help would be great

PML vs Graphic example.phxproj (1020 KB)

2

Slightly modified version of the file with more experimentation. I have three models,

  1. The original graphical model
  2. The PML version where I just hit the button and run ( So it has deriv (A1= - k*A1 etc…
  3. The PML version where I switch the amounts to concentrations (A2 → C2 which is how the PML school programed that one lesson)

Models 1 and 2 are the same but 3 is different. All that was changed was switching from amount based RHS to a concentration based RHS. Clearly I break something but it is not intuitive to me where I go wrong.

PML vs Graphic example v2.phxproj (1.33 MB)

3

Consider: CL=k*V, C=A1/V

dA1 = -(CL * C) is same as dA1 = -k*V * A1/V (Volumes cancel out)

You have: deriv(A1 = - (Cl * C)- (A1 * K_tissue_in- A2 * K_tissue_out))

You then replace A1 with C when A1 = C * V. You need to multiply by V, e.g.,

deriv(A1 = - (Cl * C)- (C* V * K_tissue_in- C2 * V2 * K_tissue_out))

Make this change and you will see the models produce the same results

It might be easier to read if you parameterize your whole model into either microconstants or clearance/volume, and not mix the 2 but that is up to the modelers preference!