A question about "set nseg for cable"

[boboppie]

Dear ted,

I'm a beginner of NEURON, and I have a question about how to find a proper nseg value for an axon.

Here the length of my passive cable is L = 100, g_pas = 0.00005. The diameter will vary between 0.01 and 1. Each segment should be shorter than lambda/20. (lambda is the electrotonic length). Others are default values.

First, according to "lambda = sqrt(r_m*diam/(4*Ra))", r_m = 1/g_pas, than I get the nseg should between 3 and 17.

I'm not sure if I get the right answer?

Please help me

Thanks :)

[ted]

The diameter will vary between 0.01 and 1.

0.01 to 1 micron? For such a wide range of diameters, it would be best to represent the
cable with several sections. Are you using morphometric cata, e.g. from Neurolucida, in
which diameter varies irregularly over the length of the axon, or is this a "theoretician's
axon" that is conceived as a series of cylinders or conical frustae?

Each segment should be shorter than lambda/20.

DC length constant is a useful guide for spatial discretization only if you are interested in
steady state (DC) signals, and don't care about signals that vary with time. Otherwise,
discretization should be based on AC length constant. Read about the d_lambda rule here:
Hines, M.L. and Carnevale, N.T. NEURON: a tool for neuroscientists.
The Neuroscientist 7:123-135, 2001.
which is available from a link at
http://www.neuron.yale.edu/neuron/bib/nrnpubs.html

Also, if you are interested in membrane potential at the midpoint of any section, be sure to
use an odd value for nseg.

[boboppie]

Dear Ted,

Thank you for replay.

My original question is to model electrically coupled neurons. Such coupling is known to exist in a number of cases, for instance between cells in the retina and between inhibitory neurons in the hippocampus. Although in biology such coupling occurs via so called gap junctions, the coupling is here modeled as a cable between the two cells. The cells are modeled as single compartments.

In Neuron I created a model consisting of two cells (soma): a large one modeled as a cylinder with a length 10 and diameter 20 μm, and a smaller one with length 10 and diameter 5μm. Connect the cells with a passive cable (dend) of 100μm length, (insert pas) but with g_pas = 0.00005S/cm2.

When I create the dend, I need to set nseg for the cable correctly, knowing that I will vary the diameter between 0.01 and 1 μm to see the spike propagation. Each segment should be shorter than lambda/20. This is usually a decent choice the get sufficient numerical accuracy.

So, as I stated in my first post, I get the value for the nseg. Do you think it is correct?

Thanks :)

[ted]

Connect the cells with a passive cable

If this is simply a cylinder, then just use the CellBuilder to construct the entire model.
Then you can use the Geometry page's d_lambda tool to take care of nseg. All problems
solved.

[boboppie]

Dear Ted,

Thank you very much. CellBuilder is great! :)