temperature/celsius and qt

NMODL and the Channel Builder.
Post Reply
NatalieD

temperature/celsius and qt

Post by NatalieD »

Hi,
Im having problems with making sure that all my MOD files are similar with regard to qt and temperature. What is the default celsius value in NEURON? I tried to set a value for celsius in a mod file - but i couldn't (when i compiled it stated that the parameter would be ignored and set by NEURON).
Would you be able to briefly explain the most efficient way to make sure that I can have some sort of agreement between all of my MOD files for temperature and qt. Thank you.
ted
Site Admin
Posts: 5793
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: temperature/celsius and qt

Post by ted »

NatalieD wrote:What is the default celsius value in NEURON?
celsius is a hoc keyword. To determine its value, start NEURON, then in the xterm at the
oc> prompt type
celsius
then press the return key.
Result: 6.3 (appropriate for the HH squid axon model).
I tried to set a value for celsius in a mod file - but i couldn't (when i compiled it stated that the parameter would be ignored and set by NEURON).
Correct. To assign a new value to celsius, use a hoc statement. Example
celsius = 100
Would you be able to briefly explain the most efficient way to make sure that I can have some sort of agreement between all of my MOD files for temperature and qt. Thank you.
In the Programmer's Reference, you will find celsius listed as one of the global variables.
http://www.neuron.yale.edu/neuron/stati ... ml#globals
This means that it has the same value throughout a model. The other globals of note are
stoprun, secondorder, dt, and t. When the local variable time step integrator is used, dt
and t are global only within a model cell (each model cell has its own dt and t).

Proper usage of celsius in
a mod file is to declare it in the ASSIGNED block
ASSIGNED {
. . . other declarations . . .
celsius (degC)
. . . other declarations . . .
}
Then its value will automatically be known to that mechanism, which can use it to
calculate a rate constant or whatever else you want to do with it.

Big caveat: celsius has no effect whatever on most mechanisms. It affects only those
that have statements that explicitly use it to calculate a rate constant etc..
MOswald
Posts: 6
Joined: Tue Nov 14, 2006 12:15 am
Location: University of Otago Medical School
Contact:

temperature dependent rate functions in mod files

Post by MOswald »

Hi Ted,

to follow up on this,

I would like to incorporate the effect of temperature on rates specified within mod files. Many mod files commonly specify a qt variable such as

qt = q10^((celsius-22 (degC))/10 (degC))

where q10 = 3 or sometimes 4.5, and the qt is then incorporated in the rate function such as in

taun = 1 / (qt*(alphan + betan))

If all my mod files have temperature dependent rate functions I can run simulations effectively at different temperatures by specifying it in the hoc file with eg:
celsius = 30

Please correct me if I am wrong.
My question is how the qt function is specified at the first place. The q10 variable seems to be a constant that is either set at 3 for ion mechanisms or at 4.5 if cyclic nucleotide activators are involved. I don't understand where the value of the subtraction from celsius comes from in this equation as it is not always the same. Does this refer to the experimental temperature at which rate constants were derived from or is it unique to each channel mechanism?
The final division by 10 seems to common to all qt functions so I assume it means that the rate changes by one unit for every 10 degree celsius step.

Perhaps my most important question comes at the end. When I adapt parameters of an existing mod file from the model database, should I just set celsius in the hoc file to that at which the experiments were conducted which I want to adapt for my specific cell type. Or alternatively is it reasonalbe to change anything in the qt function itself (eg the value of the subtraction from celsius)?

I hope this makes all sense and that you can verify this issue for me.
Manfred
ted
Site Admin
Posts: 5793
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: temperature dependent rate functions in mod files

Post by ted »

MOswald wrote:I don't understand where the value of the subtraction from celsius comes from in this equation as it is not always the same. Does this refer to the experimental temperature at which rate constants were derived
True.
The final division by 10 seems to common to all qt functions so I assume it means that the rate changes by one unit for every 10 degree celsius step.
False.
As you noted, the form of equation commonly used is
qt = q10^((celsius-temp0 (degC))/10 (degC))
where temp0 is the temperature at which the rates were experimentally determined.
Removing the units information
qt = q10^((celsius-temp0)/10)
may help you see that (celsius-temp0)/10 is the power to which q10 is raised.
So if q10 is 2, a 10 degC temperature shift makes rates twice as fast (if the new temp is
warmer).
When I adapt parameters of an existing mod file from the model database
you need to know two things: the temperature at which the rates were experimentally
measured, and the value of q10. It's not as simple as deciding "is it 3 or is it 4.5?"
In most cases, rates were determined at just one temperature and nobody knows for
sure what q10 should be. See
How to implement temperature dependent rates
https://www.neuron.yale.edu/phpBB2/view ... hlight=q10
Last edited by ted on Sun Dec 23, 2007 6:04 pm, edited 1 time in total.
MOswald
Posts: 6
Joined: Tue Nov 14, 2006 12:15 am
Location: University of Otago Medical School
Contact:

Q10 and qt

Post by MOswald »

Thank you,

That all makes a lot more sense now!

Manfred
Post Reply