www.neuron.yale.edu

Hi,

I am not sure if this falls into the realm of parellalization so I have posted this query here.
I want to know whether it is possible to simultaneously simulate a group of neurons, all of which may not necessarily be connected to each other but they do share the same extracellular space. Also, the behaviour of each neuron is different from the other in the sense that while a particular neuron is spiking rapidly, the others may be spiking at different rates. The aim being to study the effect of spiking activity in one neuron on the behavior of another due to the changes in the composition of the extracellular space.

Where I am confused is the following

(a) While I can make a template for neurons(at present I am considering a simple single compartment model) and use it to create multiple neurons, how do I lay them out in a spatial sense and make them share the same extracellular space. (For example as a starting point I want to layout 9 neurons in a 3x3 grid with extracellular space between and around the neurons as show below)

===========
= N = N = N =
===========
= N = N = N =
===========
= N = N = N =
===========

(b) How do i ensure that the ion concentration in the extracellular space is being read and written by multiple copies(one in each neuron) of the same mechanism simultaneously. This seems logically incorrect. I guess i would have to update the ion concentration due to activity of one neuron at a time and then sequentially for the rest.

(c) And in between the updation of the extracellular space by the neurons, if I want to ensure some sort of diffusion within the extracellular space based on the spatial location of the neurons, how do I do that.

Is this the correct way of looking at things.


thanks in advance

auatai wrote:I am not sure if this falls into the realm of parellalization

It does not.

I want to know whether it is possible to simultaneously simulate a group of neurons, all of which may not necessarily be connected to each other . . . the behaviour of each neuron is different from the other in the sense that while a particular neuron is spiking rapidly, the others may be spiking at different rates.

This is possible.

but they do share the same extracellular space . . .The aim being to study the effect of spiking activity in one neuron on the behavior of another due to the changes in the composition of the extracellular space.

NEURON is best suited for representations of intra- and extracellular space that take the form of (possibly multi-layer) cables or singly connected trees constructed from cables. Propagation of electrical and/or chemical signals is easy to implement in the longitudinal (along the length of a cable) and/or radial directions (perpendicular to the axis of the cable). There is no native representation of general three dimensional space, so workarounds are necessary to set up electrical or chemical coupling between compartments that are not nearest neighbors in the longitudinal or radial directions.

(a) While I can make a template for neurons(at present I am considering a simple single compartment model) and use it to create multiple neurons, how do I lay them out in a spatial sense

Templates exported by the CellBuilder and Network Builder have a public method called position(), but this only alters the location of the image that is rendered in a shape plot. It has no effect on simulation results.

An important aside: single compartment models cannot affect, or be affected by, extracellular potential.

make them share the same extracellular space. (For example as a starting point I want to layout 9 neurons in a 3x3 grid with extracellular space between and around the neurons as show below)

I'd think about trying to couple NEURON to some other simulator that is capable of solving diffusion problems, and let each do what it does best and pass results to the other. I hesitate to speculate further, lacking any knowledge about the availability and capabilities of such simulators, but given NEURON's ability to use Python code, and the rapid growth of Python in scientific programming, that's where I'd suggest you look first.

Thanks Ted,

I'll try to do it some other way