Prev: Recognizing a key hit while running
Next: Is there a manual on reading Matlab segmentation fault stack traces?
From: Pascal Schulthess on 28 Apr 2010 05:06
Hi everyone,

I recently started playing around with the SimBiology Toolbox. Although it tykes some time to get used to, I really like it. Especially the sensitivity analysis functionality. What I miss though, is the possibility to calculate sensitivities with respect to reaction rates. In metabolic control analysis this is reflected in the control and elasticity coefficients.

Perhaps I missed this, but can anyone help me with that.

Thanks in advance,

Pascal
From: Pascal Schulthess on 28 Apr 2010 07:07
"Pascal Schulthess" <hello(a)pascalschulthess.de> wrote in message <hr8tpt$4k3$1(a)fred.mathworks.com>...
> Hi everyone,
>
> I recently started playing around with the SimBiology Toolbox. Although it tykes some time to get used to, I really like it. Especially the sensitivity analysis functionality. What I miss though, is the possibility to calculate sensitivities with respect to reaction rates. In metabolic control analysis this is reflected in the control and elasticity coefficients.
>
> Perhaps I missed this, but can anyone help me with that.
>
> Thanks in advance,
>
> Pascal

Hi again,

I just noticed that I forgot something in previous post. Is there an "easy" way to calculate the (local/global) response matrix with the SimBiology Toolbox?

Thanks a lot,

Pascal
From: Pramod on 28 Apr 2010 17:09
Hello Pascal,

From your mail it seems that you would like to calculate sesitivities of the reaction fluxes. This is currently not possible in SimBiology.

However, You can calculate the sensitivities of the species and use that data to calculate the sensitivities of the reaction fluxes.

For example if you have a reaction R1 whose rate = k1*x*y then

flux = f = k1*x*y

Sens(f) = df/dp = d(k1*x*y)/dp = k1*dx/dp*y + k1*x*dy/dp = k1*(dx/dp * y + x * dy/dp) = k1(Sens(x)*y + Sens(y)*x)

Where p is a parameter, and Sens(x) mean s sesitivity of x with respect to p.

Let me know if you have further questions.

Are you using SimBiology desktop or command line?

Can you clarify what you meant by response matrix.

- Pramod
From: Pascal Schulthess on 7 May 2010 09:33
Hi Pramod,

thanks for the help and sorry for my late answer.

What I meant by is response coefficient is the sensitivity of a species to small perturbations in another species (dx/dy). I discovered that I can calculate this with:

spe = sbioselect(model,'Type','species');
set(configsModel.SensitivityAnalysisOptions,'SpeciesInputFactors', spe);
set(configsModel.SensitivityAnalysisOptions,'SpeciesOutputs',spe);
simModel = sbiosimulate(model,configsModel,[],[]);
[T, R, snames, ifacs] = getsensmatrix(simModel);
R2 = squeeze(R);

This gives my the continuous sensitivity of every species to every other species. But now I'm struggling with the steady state response. Is there a way to calculate R2 at steady state?

Thank so much!

Pascal

"Pramod " <pramod.kumar(a)mathworks.com> wrote in message <hra85h$pmh$1(a)fred.mathworks.com>...
> Hello Pascal,
>
> From your mail it seems that you would like to calculate sesitivities of the reaction fluxes. This is currently not possible in SimBiology.
>
> However, You can calculate the sensitivities of the species and use that data to calculate the sensitivities of the reaction fluxes.
>
> For example if you have a reaction R1 whose rate = k1*x*y then
>
> flux = f = k1*x*y
>
> Sens(f) = df/dp = d(k1*x*y)/dp = k1*dx/dp*y + k1*x*dy/dp = k1*(dx/dp * y + x * dy/dp) = k1(Sens(x)*y + Sens(y)*x)
>
> Where p is a parameter, and Sens(x) mean s sesitivity of x with respect to p.
>
> Let me know if you have further questions.
>
> Are you using SimBiology desktop or command line?
>
> Can you clarify what you meant by response matrix.
>
> - Pramod
From: Arthur Goldsipe on 7 May 2010 14:42
Hi Pascal,

I'd like to clarify how SimBiology uses 'SpeciesInputFactors', and make sure this is really what you mean by response coefficient. If you specify x as a SpeciesOutput and y as a SpeciesInputFactor, then SimBiology will calculate the sensitivity of species x (at some time t) with respect to the *initial amount* of species y. Hopefully, this is what you mean, since a perturbation in species y at time t shouldn't affect x at time t, but I just wanted to make sure.

Unfortunately, SimBiology does not yet calculate steady states. (In general, a model could be oscillatory or chaotic and so not even have steady states!) However, we know that in many cases steady states are well-defined, and this is definitely a feature we've been thinking about for the future.

In the interim, I think your best option is to set the stop time of the simulation to something that will approximate your steady state. Then, the approximate steady-state sensitivities will be the final row of the matrix R (where R is the second output from getsensmatrix).

Good luck!
-Arthur

"Pascal Schulthess" <hello(a)pascalschulthess.de> wrote in message <hs14qg$6pn$1(a)fred.mathworks.com>...
> Hi Pramod,
>
> thanks for the help and sorry for my late answer.
>
> What I meant by is response coefficient is the sensitivity of a species to small perturbations in another species (dx/dy). I discovered that I can calculate this with:
>
> spe = sbioselect(model,'Type','species');
> set(configsModel.SensitivityAnalysisOptions,'SpeciesInputFactors', spe);
> set(configsModel.SensitivityAnalysisOptions,'SpeciesOutputs',spe);
> simModel = sbiosimulate(model,configsModel,[],[]);
> [T, R, snames, ifacs] = getsensmatrix(simModel);
> R2 = squeeze(R);
>
> This gives my the continuous sensitivity of every species to every other species. But now I'm struggling with the steady state response. Is there a way to calculate R2 at steady state?
>
> Thank so much!
>
> Pascal
 |  Next  |  Last
Pages: 1 2
Prev: Recognizing a key hit while running
Next: Is there a manual on reading Matlab segmentation fault stack traces?