synapses seem to have no influence

Moderator: wwlytton

Post Reply
srene
Posts: 14
Joined: Thu Jun 27, 2013 3:46 pm

synapses seem to have no influence

Post by srene » Wed Apr 30, 2014 9:04 am

Hello,

Iḿ working on a network model of cells of one template. Each cell is connected by gaps and synapses. Then a part of the cells will be stimulated in order to make the whole network firing synchronously. The way of connecting with synapses is shown in the code above:

Code: Select all

nZellen = 100 // number of cells in network
syndel = 1      // synaptic delay

for i = 0, nZellen-1 {
	Zellen[i] = new singing_fish_cell()  // singing_fish_cell() is the template of one cell one cell has 14 dends thats why k is from 0 to 14
}
proc Synapsen() {local startZelle, endZelle,  k, d
	
	//cancel all synapses buillt before and start a new connection
	for j = 0, nZellen-1 {
			Zellen[j].nclist.remove_all()
	}

   // theese vectors for putting them in a  .dat file

syn_start = new Vector()
syn_end = new Vector()              
dend_end = new Vector()
syn_del = new Vector()    

   // til here

	//define start and end of synapse
	for i = 0,nSynapsen-1 {
		startZelle = int(r.uniform(0,nZellen))
		endZelle = int(r.uniform(0,nZellen))
		k = int(r.uniform(0,14))             // random value for the dends
		d = int(r.uniform(0,syndel))       //  if syndel is set as 1 d will always be 0
				
		if (startZelle == endZelle) {
			endZelle = endZelle + 1
		}
		if (endZelle >= nZellen) {
			endZelle = 0
		}
		
// Using NetCon and building synapses with random values
		Zellen[endZelle].dend[k] syn[i] = new ExpSyn(0)
		Zellen[startZelle].axon Zellen[endZelle].nclist.append(new NetCon(&v(1), syn[i], threshold, d, weight))

// writing the random numbers in the vectors for .dat file
	syn_start.append(startZelle)	
	syn_end.append(endZelle)
	dend_end.append(k)
	syn_del.append(d)	
// til here
	}
}
If syndel is set as 1 the random value d is always zero, so the synaptic delay is always zero. The network works with gaps (means a part is stimulated and other cell which are not simulated do also fire). But without setting gaps it doesn´t work even if syndel is set as 1. The synapses should have no delay. In my .dat file syn_del is always zero so there should be a influence of synapses but even if I plot the current of the synapses using GUI the current is zero.

Can you imagine what is going on and why the synapses seem to have no influence in this network.

René

ted
Site Admin
Posts: 5057
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: synapses seem to have no influence

Post by ted » Wed Apr 30, 2014 10:45 am

The first question is: what synaptic connections actually exist in your model?
The second question is: what are the weights of the NetCons that establish these connections?

So you need to discover the actual connectivity of your network. This means generating a list that reports each synaptic connection as
identity of presynaptic cell, identity of postsynapic cell, identity of synaptic mechanism attached to postsynaptic cell
This paper shows how. Its source code is available from ModelDB. Note that Lists are used for all collections of objects; this is actually more practical than using "subscripted object reference names."
Hines, M.L. and Carnevale, N.T.
Translating network models to parallel hardware in NEURON.
J. Neurosci. Methods 169:425-455, 2008.

This may all seem difficult if not tedious, but in the long run it will make things easier.

srene
Posts: 14
Joined: Thu Jun 27, 2013 3:46 pm

Re: synapses seem to have no influence

Post by srene » Thu May 01, 2014 1:32 pm

Sorry that I didn´t mention that. For the weight of the synapses I varied the value of course.

I tried following weights:

Code: Select all

weight = 0.7
weight = 0.9
weight = 10
weight = 100
Now the last two values may be unrealistic but It was on purpose. I think the there exists synapses resp. there is connectivity because if I choose that my network of 25 cells(nZellen= 25) should be connected by 200 synapses (nSynapsen = 200) and run the simu, I can choose at the GUI:
Neuron Main Menu -> Graph -> Current axis -> rightclick on Graph window -> Plot what? -> show -> All-> syn[0].i until syn[199].i or I can choose syn[0-199].e -.g -.loc -.get_loc -.has_loc -.tau

Is this a proof that the network is connected? Or can I proof It another way?

René

ted
Site Admin
Posts: 5057
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: synapses seem to have no influence

Post by ted » Thu May 01, 2014 4:12 pm

You haven't answered the key questions:
What synaptic connections actually exist in your model?
What are the weights of the NetCons that establish these connections?
Is this a proof that the network is connected?
No. The answers to these questions exist in the computer after you execute the hoc code that sets up your model. The paper I cited shows how to discover the synaptic connections that actually exist, and you can download the source code used in that paper from ModelDB. It would be very easy to modify proc tracenet() so that it also reports the NetCon weights.

As for the weights you tried, are those just guesses, or have you actually tried activating a synapse with an event associated with one of those weights, and seen what it does to the postsynaptic cell?

Post Reply