How do I fit batch data to a simbiology model?

I have a dataset that contains one independent variable, time and four dependent concentration variables. These are kinetic traces. Two of these variables are titrated concentrations that change from one trace to the next. I would like to fit the entire dataset to my model however Simbiology does not allow datasets where the independent variable increases non monotonically. How do I fit the entire dataset globally?

Attached is a 3D plot of the different traces. I would like to create surface with my model on it. Is this possible in simbiology. I suppose if I get the parameters I can do it manually using meshgrid and surf. However the parameters need to come from a global fit. There are no analytical solutions to the 6 set of coupled ODEs. It has to be done numerically.

I have also attached the dataset.

Answer 1

[Jeremy Huard]

Hi,

you will need first to assign a unique ID to every experiment. It looks like T7 is fixed during every experiment, is this correct?

If this is the case, you can use T7 as ID. If not, you can create new IDs with findgroups using the set of parameters you vary manually.

After setting the new variable (e.g. ID) as grouping variable, you can run a fit program as pooled fit to estimate a unique set of parameter values to fit all experiments.

Best,

Jérémy

[Jeremy Huard]

Oh, and one more thing:

if experimental conditions for every ID differ (e.g. T7 corresponds to a parameter in the model), you can assign the value of T7 for each ID by adding a row Variant from data to the Data map section of a fit program.

[Archishman Ghosh]

A little more detailed assistance would be appreciated. If I change the T7 colum as a group how to I use the values as inputs for the T7 parameter?

[Jeremy Huard]

This can be done in a fit program, not in the data definition itself. You can assign the value of T7 for each ID by adding a row Variant from data to the Data map section of a fit program:

[Archishman Ghosh]

Thanks a lot. I think the issue was that I was using the 2021a version which did not have this data map option.

I managed to make T7 and RNase a variant from data with a serparate group column added as 1,1,1,1,1,....2,2,2,2,....3,3,3,... etc. which seemed to work in taking each experiment as a separate group.

However, I can't seem to make the variant group specific. May I ask what's the trick to do this?

As it appears the fits are not doing anything group specific. Perhaps it just averaged all of them? The fits are the same for the 5 different experimental conditions. Is this because I didn't do a group specific variant?

Let me know if you would like to have a look at the model as well.

best,

Archie

[Jeremy Huard]

Hi @Archishman Ghosh,

you are correct. Fitting with group-specific variants is only supported in R2021b or newer releases.

The Variant and dose setup table in the second screenshot you shared looks correct. The mention Data below variant2 and variant3 means that these variants are extracted from the dataset. Other columns with the mention Group Specific are there for you to select variants stored in the model. These columns are editable.

This definition is admittedly a bit confusing because variants extracted from the dataset are also group specific.

So far, your fitting setup seems correct. It would be useful if you could share the project with me to investigate further.

Could you please write to me via my user profile to continue this thread over email?

Thanks,

Jérémy

[Archishman Ghosh]

Thanks, I've sent you a personal message.