www.neuron.yale.edu

Hello,

for my master thesis i use some neurons defined with hoc-files. I am determining the surface area of all sections, inclusive the soma with the following formula:

A=area(.5)*nseg

For a particular population of neurons the length L of the soma is only L=0.01. So NEURON calculates an surface area for the soma being very small.

So my question is: Is it correct, that NEURON uses always a cylinder-model for the soma for the simulation? If not, does NEURON provide a function for calculating the surface area of the soma represented by a ball?

greetings, Franzi

Franzi wrote:I am determining the surface area of all sections, inclusive the soma with the following formula:

A=area(.5)*nseg

Guaranteed to produce nonsense results if diam is not constant. Do this instead:

For a particular population of neurons the length L of the soma is only L=0.01. So NEURON calculates an surface area for the soma being very small.

Several important points here.
1. An anatomist can look at a cell and tell you what its soma is. NEURON can't. If you
create a ball and stick model that has a 20x20 um section you call "axon", and another
section with 1 um diam and 1000 um length that you call "soma", an anatomist would
give you a dirty look, but NEURON will tell you that the "soma" has an area of 1000*PI
square microns.
2. An anatomical soma may be represented in a model by multiple sections. This commonly
happens with models based on morphometric data when the operator interrupts a sequence
of somatic measurements with measurements of axons and dendrites that are attached
to the soma. Total soma area would then be the sum of the areas of each of the soma
sections.
3. Some programs that convert morphometric data files to hoc code introduce a bogus
"soma" section as a root for the branched architecture of the cell. cvapp does this. The
bogus section, which is called soma, has an extremely short length so it doesn't disturb
electrical or chemical signaling properties of the model. Needless to say, the area of the
bogus soma is completely meaningless.

So my question is: Is it correct, that NEURON uses always a cylinder-model for the soma for the simulation?

False.

If not, does NEURON provide a function for calculating the surface area of the soma represented by a ball?

The func presented above will report the true section area regardless of morphology.
If you have the misfortune to work with morphometric data that results in a fragmented
soma (multiple sections represent the anatomical soma), or hoc files created by cvapp
which have a bogus soma section, you'll have to deal with it yourself by figuring out which
sections constitute the anatomical soma and adding their areas.