Sensitivity analysis using COPASI
577 views
Skip to first unread message
Quan Zhou
Mar 15, 2016, 1:21:39 PM3/15/16
to COPASI User Forum
Hi all,
I am working on the sensitivity analysis for an Acetone-butanol-ethanol fermentation process. Previously I used parameter estimation to estimate of all the parameters in the pathway( please see attached picture), now I would like to use the sensitivity analysis to calculate the change in butanol production in response to a 5% increase in each parameter in the model. So...what should I choose for the subtask? in the effect blank, should I choose B (which stand for butanol)? and what is the cause? and What is the delta factor? should I choose 0.05 as 5% increase in each parameter in this case?
Thanks,
Quan.
Stefan Hoops
Mar 24, 2016, 8:56:18 AM3/24/16
to copasi-u...@googlegroups.com
Hello Quan,
Sensitivity of a value Y with respect to the variable X is the defined
as dX/dY, i.e., it is an infinitesimal change of X and Y. This
derivative is numerically estimated in COPASI. If you need the
sensitivity with respect to a non infinitesimal change in X you can use
the scan and calculate 3 values for X in the interval [0.95*X, 1.05*X].
Thanks,
Stefan
--
Stefan Hoops, Ph.D.
Research Assistant Professor
Director, Biochemical Networks Modeling Group
Center for Modeling Immunity to Enteric Pathogens (MIEP)
Faculty of Health Sciences (FHS)
Biocomplexity Institute of Virginia Tech
1015 Life Science Circle (0477)
Blacksburg, Va 24061, USA
Phone: (540) 231-1799
Fax: (540) 231-2606
Email: sho...@vbi.vt.edu
Sensitivity of a value Y with respect to the variable X is the defined
as dX/dY, i.e., it is an infinitesimal change of X and Y. This
derivative is numerically estimated in COPASI. If you need the
sensitivity with respect to a non infinitesimal change in X you can use
the scan and calculate 3 values for X in the interval [0.95*X, 1.05*X].
Thanks,
Stefan
Stefan Hoops, Ph.D.
Research Assistant Professor
Director, Biochemical Networks Modeling Group
Center for Modeling Immunity to Enteric Pathogens (MIEP)
Faculty of Health Sciences (FHS)
Biocomplexity Institute of Virginia Tech
1015 Life Science Circle (0477)
Blacksburg, Va 24061, USA
Phone: (540) 231-1799
Fax: (540) 231-2606
Email: sho...@vbi.vt.edu
Pedro Mendes
Mar 24, 2016, 10:03:38 AM3/24/16
to copasi-u...@googlegroups.com
Dear Quan,
just to expand on what Stefan described: Sensitivity analysis is indeed
infinitesimal, however since 5% is still a small change you can use the
results of sensitivity to give you a feeling of which parameter has the
strongest effect on butanol. The result of sensitivities indicates what
would be the change in butanol for 1% change in the parameters (so a
*draft* estimate would be that time 5, but note that this is an
approximation and it would not be the correct value).
With parameter scan you can indeed find out the exact value of the
changes in butanol with 5% change in each parameter. For that
essentially you change one parameter by 5% and see what is now butanol.
Parameter scan allows you to do all the changes in one go. For this you
would chose time course or steady state in the subtask for parameter
scan (I prefer to look at steady state values, but time course is ok
too, depends on what your question, of course). Then add each one of the
parameters as a "scan" and set each one to have the minimum value to be
equal to what you have it in the model, the maximum value to be that
value + 5%, and finally set the number of intervals to 2. You also need
to create a report file that contains the parameter values and the
steady state concentration of butanol (the variable you care about).
When you run this you will get the results you want, but notice that you
will also get a lot more combinations of multiple parameters changed. so
look for the lines that have only one parameter changed by 5%.
Now, looking at your model (that you sent me in a private email), I see
that your model is closed and most of the fluxes are zero (of course,
because there is no force maintaining the steady state). This results
that there is no increase in butanol since its value is at equilibrium
no matter the constants... I suggest that you should look at your model
and set the pathway substrate and final product to be constant (rather
than depend on reactions), otherwise you will not get any meaningful
results out of this model.
Pedro
Pedro Mendes
Professor of Computational Systems Biology
School of Computer Science
Manchester Centre for Integrative Systems Biology
University of Manchester
Manchester Institute of Biotechnology
131 Princess Street
Manchester, M1 7DN, U.K.
just to expand on what Stefan described: Sensitivity analysis is indeed
infinitesimal, however since 5% is still a small change you can use the
results of sensitivity to give you a feeling of which parameter has the
strongest effect on butanol. The result of sensitivities indicates what
would be the change in butanol for 1% change in the parameters (so a
*draft* estimate would be that time 5, but note that this is an
approximation and it would not be the correct value).
With parameter scan you can indeed find out the exact value of the
changes in butanol with 5% change in each parameter. For that
essentially you change one parameter by 5% and see what is now butanol.
Parameter scan allows you to do all the changes in one go. For this you
would chose time course or steady state in the subtask for parameter
scan (I prefer to look at steady state values, but time course is ok
too, depends on what your question, of course). Then add each one of the
parameters as a "scan" and set each one to have the minimum value to be
equal to what you have it in the model, the maximum value to be that
value + 5%, and finally set the number of intervals to 2. You also need
to create a report file that contains the parameter values and the
steady state concentration of butanol (the variable you care about).
When you run this you will get the results you want, but notice that you
will also get a lot more combinations of multiple parameters changed. so
look for the lines that have only one parameter changed by 5%.
Now, looking at your model (that you sent me in a private email), I see
that your model is closed and most of the fluxes are zero (of course,
because there is no force maintaining the steady state). This results
that there is no increase in butanol since its value is at equilibrium
no matter the constants... I suggest that you should look at your model
and set the pathway substrate and final product to be constant (rather
than depend on reactions), otherwise you will not get any meaningful
results out of this model.
Pedro
Professor of Computational Systems Biology
School of Computer Science
Manchester Centre for Integrative Systems Biology
University of Manchester
Manchester Institute of Biotechnology
131 Princess Street
Manchester, M1 7DN, U.K.
Quan Zhou
Mar 24, 2016, 1:27:24 PM3/24/16
to COPASI User Forum
Hi Pedro,
Thanks for all your information and correction for my model.
I think I will correct my model first and then try the parameter scan like you and stefan suggest me to do. But before I correct my model, would you please help me make sure that I am understanding correctly for your suggestion?
Like the attached picture, is that correct if change the simulation type of all the substrates and final products to be fixed not reactions? And is that correct if I type in the initial value for each substrate under the initial concentration tab? ALSO, for the product (butanol, acetone..etc) , where should I type in my final products concentration?
Thanks,
Quan.
在 2016年3月24日星期四 UTC-4上午10:03:38，pedro.mendes写道：
在 2016年3月24日星期四 UTC-4上午10:03:38，pedro.mendes写道：
Pedro Mendes
Mar 24, 2016, 4:12:24 PM3/24/16
to copasi-u...@googlegroups.com
Quan,
no, you should not set all species to fixed. If you do this then there
will be no model at all, just constants. You should only set to fixed
the entry points in the pathway and the exits. I am not sure what these
should be, but looking at a quick diagram of your reactions, I think
these are most likely:
F6P (since you don't have glucose, this looks like the entry of carbon)
lactate (obviously a product)
acetate (not sure about this one, but it is not made from anything in
your pathway, so likely a substrate too)
inactivecell (this is clearly an end product)
A (only has reactions producing it, so it will accumulate if you don't
fix it)
B (same as A)
E (same as A)
Acet (only has reactions consuming it, so it must be a substrate)
X (only has reactions consuming it, so it must be a substrate)
But, of course, I don't know much about your model...
Pedro
> --
> You received this message because you are subscribed to the Google
> Groups "COPASI User Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-fo...@googlegroups.com>.
> To post to this group, send email to copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com
> <https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.
no, you should not set all species to fixed. If you do this then there
will be no model at all, just constants. You should only set to fixed
the entry points in the pathway and the exits. I am not sure what these
should be, but looking at a quick diagram of your reactions, I think
these are most likely:
F6P (since you don't have glucose, this looks like the entry of carbon)
lactate (obviously a product)
acetate (not sure about this one, but it is not made from anything in
your pathway, so likely a substrate too)
inactivecell (this is clearly an end product)
A (only has reactions producing it, so it will accumulate if you don't
fix it)
B (same as A)
E (same as A)
Acet (only has reactions consuming it, so it must be a substrate)
X (only has reactions consuming it, so it must be a substrate)
But, of course, I don't know much about your model...
Pedro
> You received this message because you are subscribed to the Google
> Groups "COPASI User Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-fo...@googlegroups.com>.
> To post to this group, send email to copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com
> <https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.
Quan Zhou
Mar 24, 2016, 4:32:49 PM3/24/16
to copasi-u...@googlegroups.com
Dear Pedro,
I am so sorry for misleading you for my question. Please bear me to clarify this. My question is if I fix my entry point (in my case, just xylose as the carbon source), should I choose the 'fixed" to be the simulation type? And do I need to type in the initial concentration as 259.59 mmol/L in my model like shown in the attached picture? Also, do I need to type in the final concentration of xylose from my experiment? if yes, where should i I put the final concentration of xylose. In addition, My products are lactate, acetate, butyrate, acetone (short for A), ethanol (short for E) and Butanol (short for B), how should I fix those values in the model? I just saw an initial concentration tab but I did not see a final concentration tab under 'species" tab. Please help me with this. I also attached my copasi file with the model pathway.
Thanks,
Quan.
--
You received this message because you are subscribed to a topic in the Google Groups "COPASI User Forum" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/copasi-user-forum/RiteXY1XloY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to copasi-user-fo...@googlegroups.com.
To post to this group, send email to copasi-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/copasi-user-forum/56F44A25.7080100%40manchester.ac.uk.
Quan Zhou
Graduate Student
Biological and Agricultural Engineering
Room 120, David S. Weaver Laboratories
North Carolina State University
Raleigh, NC 27695-7625
Cell: 919-592-3935
Email: qzh...@ncsu.edu
Pedro Mendes
Mar 24, 2016, 4:47:34 PM3/24/16
to copasi-u...@googlegroups.com
Dear Quan,
there is no "final concentration" for species that are fixed. If they
are fixed it means that their concentration does not change (so the
final is the same as the initial). If xylose is your carbon source but
it it changes its concentration along time, then your model is more
complex than normal (or you will have to extend it upstream of xylose).
You need to make the model match what the conditions of the experiment
(whether it is a real experiment or a conceptual one).
But to answer your more direct question, yes you set a species to be
fixed in the way you showed in your image.
Pedro
On 03/24/2016 04:32 PM, Quan Zhou wrote:
> Dear Pedro,
>
> I am so sorry for misleading you for my question. Please bear me to
> clarify this. My question is if I fix my entry point (in my case, just
> xylose as the carbon source), should I choose the 'fixed" to be the
> simulation type? And do I need to type in the initial concentration
> as 259.59 mmol/L in my model like shown in the attached picture? Also,
> do I need to type in the final concentration of xylose from my
> experiment? if yes, where should i I put the final concentration of
> xylose. In addition, My products are lactate, acetate, butyrate,
> acetone (short for A), ethanol (short for E) and Butanol (short for B),
> how should I fix those values in the model? I just saw an initial
> concentration tab but I did not see a final concentration tab under
> 'species" tab. Please help me with this. I also attached my copasi file
> with the model pathway.
>
> Thanks,
>
> Quan.
>
> On Thu, Mar 24, 2016 at 4:12 PM, Pedro Mendes
> <pedro....@manchester.ac.uk <mailto:pedro....@manchester.ac.uk>>
> > Quan Zhou <qzh...@ncsu.edu <mailto:qzh...@ncsu.edu>
> <mailto:copasi-user-forum%2Bunsu...@googlegroups.com>
> <mailto:copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-forum%2Bunsu...@googlegroups.com>>.
> <https://groups.google.com/d/msgid/copasi-user-forum/CAJB9j98VQaJistWigcq8CQ95cgaH19q6M9D2TSu4kBO34g1k%3DQ%40mail.gmail.com?utm_medium=email&utm_source=footer>.
there is no "final concentration" for species that are fixed. If they
are fixed it means that their concentration does not change (so the
final is the same as the initial). If xylose is your carbon source but
it it changes its concentration along time, then your model is more
complex than normal (or you will have to extend it upstream of xylose).
You need to make the model match what the conditions of the experiment
(whether it is a real experiment or a conceptual one).
But to answer your more direct question, yes you set a species to be
fixed in the way you showed in your image.
Pedro
On 03/24/2016 04:32 PM, Quan Zhou wrote:
> Dear Pedro,
>
> I am so sorry for misleading you for my question. Please bear me to
> clarify this. My question is if I fix my entry point (in my case, just
> xylose as the carbon source), should I choose the 'fixed" to be the
> simulation type? And do I need to type in the initial concentration
> as 259.59 mmol/L in my model like shown in the attached picture? Also,
> do I need to type in the final concentration of xylose from my
> experiment? if yes, where should i I put the final concentration of
> xylose. In addition, My products are lactate, acetate, butyrate,
> acetone (short for A), ethanol (short for E) and Butanol (short for B),
> how should I fix those values in the model? I just saw an initial
> concentration tab but I did not see a final concentration tab under
> 'species" tab. Please help me with this. I also attached my copasi file
> with the model pathway.
>
> Thanks,
>
> Quan.
>
> On Thu, Mar 24, 2016 at 4:12 PM, Pedro Mendes
> <mailto:copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-forum%2Bunsu...@googlegroups.com>>.
> To post to this group, send email to
> copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>
> <mailto:copasi-u...@googlegroups.com
> copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>
> <mailto:copasi-u...@googlegroups.com>>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com
> <https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> Pedro Mendes
> Professor of Computational Systems Biology
> School of Computer Science
> Manchester Centre for Integrative Systems Biology
> University of Manchester
>
> Manchester Institute of Biotechnology
> 131 Princess Street
> Manchester, M1 7DN, U.K.
>
> --
> You received this message because you are subscribed to a topic in
> the Google Groups "COPASI User Forum" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/copasi-user-forum/RiteXY1XloY/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-forum%2Bunsu...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com
> <https://groups.google.com/d/msgid/copasi-user-forum/85f8f1f1-2613-4fe8-8787-45ef3ae6fe1a%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> Pedro Mendes
> Professor of Computational Systems Biology
> School of Computer Science
> Manchester Centre for Integrative Systems Biology
> University of Manchester
>
> Manchester Institute of Biotechnology
> 131 Princess Street
> Manchester, M1 7DN, U.K.
>
> --
> You received this message because you are subscribed to a topic in
> the Google Groups "COPASI User Forum" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/copasi-user-forum/RiteXY1XloY/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> copasi-user-fo...@googlegroups.com
> To post to this group, send email to
> copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/56F44A25.7080100%40manchester.ac.uk.
>
> For more options, visit https://groups.google.com/d/optout.
>
>
>
>
> --
> Quan Zhou
> Graduate Student
> Biological and Agricultural Engineering
> Room 120, David S. Weaver Laboratories
> North Carolina State University
>
> Raleigh, NC 27695-7625
>
> Cell: 919-592-3935
> Email: qzh...@ncsu.edu <mailto:qzh...@ncsu.edu>
>
> For more options, visit https://groups.google.com/d/optout.
>
>
>
>
> --
> Quan Zhou
> Graduate Student
> Biological and Agricultural Engineering
> Room 120, David S. Weaver Laboratories
> North Carolina State University
>
> Raleigh, NC 27695-7625
>
> Cell: 919-592-3935
>
> --
> You received this message because you are subscribed to the Google
> Groups "COPASI User Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-fo...@googlegroups.com>.
> To post to this group, send email to copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/copasi-user-forum/CAJB9j98VQaJistWigcq8CQ95cgaH19q6M9D2TSu4kBO34g1k%3DQ%40mail.gmail.com
> --
> You received this message because you are subscribed to the Google
> Groups "COPASI User Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to copasi-user-fo...@googlegroups.com
> <mailto:copasi-user-fo...@googlegroups.com>.
> To post to this group, send email to copasi-u...@googlegroups.com
> <mailto:copasi-u...@googlegroups.com>.
> To view this discussion on the web visit
> <https://groups.google.com/d/msgid/copasi-user-forum/CAJB9j98VQaJistWigcq8CQ95cgaH19q6M9D2TSu4kBO34g1k%3DQ%40mail.gmail.com?utm_medium=email&utm_source=footer>.
Quan Zhou
Mar 24, 2016, 5:05:29 PM3/24/16
to copasi-u...@googlegroups.com
Dear Pedro,
Yes, my xylose did change along time. It decreased along time since it is a substrate, and my products (lactate, butyrate, acetate, acetone, ethanol, butanol) increased along time. So..... What should I do to modify my model? Please see my data file.
Thanks,
Quan.
To unsubscribe from this group and all its topics, send an email to copasi-user-fo...@googlegroups.com.
To post to this group, send email to copasi-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/copasi-user-forum/56F45260.8070804%40manchester.ac.uk.
Quan Zhou
Apr 5, 2016, 3:54:27 PM4/5/16
to copasi-u...@googlegroups.com
Hi Dear Professor Pedro,
Would you please help me to take a look at my model and see why I did not get any meaningful result from parameter scan using COPASI? I post my question in the copasi user forum. Thank you so much.
Best,
Quan.
To unsubscribe from this group and all its topics, send an email to copasi-user-fo...@googlegroups.com.
To post to this group, send email to copasi-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/copasi-user-forum/56F45260.8070804%40manchester.ac.uk.
Stefan Hoops
Apr 5, 2016, 6:13:30 PM4/5/16
to copasi-u...@googlegroups.com
Hello Quan,
Have a look at your reactions R10, R15, R20 if any of the modifiers goes
above a threshold the function results in NaN (not a number) since
non-integer exponents for negative values are not defined on the real
numbers. I have fixed that in the attached file.
Enjoy,
Stefan
> > https://groups.google.com/d/msgid/copasi-user-forum/56F45260.8070804%40manchester.ac.uk .
Have a look at your reactions R10, R15, R20 if any of the modifiers goes
above a threshold the function results in NaN (not a number) since
non-integer exponents for negative values are not defined on the real
numbers. I have fixed that in the attached file.
Enjoy,
Stefan
Mariangela Corso
Feb 25, 2018, 2:45:30 AM2/25/18
to COPASI User Forum
Hi,
I wrote some days ago but I didn't receive any answers. I have the same problem, I have to set sensitivity on Copasi but I don't know how I can do it. I'm trying but it's incorrect
Thanks
Mariangela
I wrote some days ago but I didn't receive any answers. I have the same problem, I have to set sensitivity on Copasi but I don't know how I can do it. I'm trying but it's incorrect
Thanks
Mariangela
Stefan Hoops
Feb 26, 2018, 10:38:34 AM2/26/18
to COPASI User Forum
Hello Mariangela,
Have you looked at our documentation:
http://copasi.org/Support/User_Manual/Tasks/Sensitivity_Analysis/
If you still have questions on how to perform sensitivity analysis in
COPASI please ask a more specific question especially on what is
incorrect with the result you obtain.
Thanks,
Stefan
Have you looked at our documentation:
http://copasi.org/Support/User_Manual/Tasks/Sensitivity_Analysis/
If you still have questions on how to perform sensitivity analysis in
COPASI please ask a more specific question especially on what is
incorrect with the result you obtain.
Thanks,
Stefan
--
Stefan Hoops, Ph.D.
Research Assistant Professor
Director, Biochemical Networks Modeling Group
Stefan Hoops, Ph.D.
Research Assistant Professor
Director, Biochemical Networks Modeling Group
Faculty of Health Sciences (FHS)
Biocomplexity Institute of Virginia Tech
1015 Life Science Circle (0477)
Blacksburg, Va 24061, USA
Phone: (540) 231-1799
Fax: (540) 231-2606
Email: sho...@vt.edu
Biocomplexity Institute of Virginia Tech
1015 Life Science Circle (0477)
Blacksburg, Va 24061, USA
Phone: (540) 231-1799
Fax: (540) 231-2606
Reply all
Reply to author
Forward
0 new messages