measuring total synaptic current

[Eleftheria Pissadaki]

Dear Forum,

I would like to ask whether or not is possible to obtain the sum of the
synaptic current carried out by a certain receptor type.

For example, may I measure the total current carried out by several
GABAa, or NMDA receptors at the cell body?

Thanks in advance

Eleftheria

[ted]

may I measure the total current carried out by several
GABAa, or NMDA receptors at the cell body?

Of course. Depending on what you need, the implementation is "extremely easy" or
merely "straightforward."

The following assumes that "receptors" means "synaptic mechanisms."

If your synaptic mechanisms use the event delivery system (i.e. contain a
NET_RECEIVE block), all synapses that have the same equilibrium potential and
time course that are attached to the same node of a model cell can be represented
by a single point process. Somas are almost invariably "isopotential" so your somatic
synapses can all be attached to the same node and can thus be represented with only
one instance each of a gabaergic and an ampaergic mechanism (easily done with
Exp2Syn and ExpSyn, using appropriate e and taus).

If synapses are scattered over multiple nodes, you'll just have to keep track of them
when you set them up, and compute a "sum on the fly" as the simulation advances.

If you need more details about either of these strategies, just ask.

[cafischer]

Hi,

I want to measure the current from all the synapses I inserted into a neuron. How can I do that?
I tried to record from all synapses in Python by:

Code: Select all

vec_i = [0]*5
for i in np.arange(0,n_syn):
    vec_i[i] = h.Vector()
    vec_i[i].record(h.AMPAsyn[i]._ref_i)

But when I run a simulation the program dies without giving an error.

Is there another way to do it. Maybe using a mod file which simplifys this?
I think you mentioned this already in your comment:

If synapses are scattered over multiple nodes, you'll just have to keep track of them
when you set them up, and compute a "sum on the fly" as the simulation advances.

Can you elaborate on that or send some code?

[ted]

Wow, everything old is new again. Or at least some old things are new. Thank you, Python!

Is there another way to do it.

There are many ways that would work.

Maybe using a mod file which simplifys this?

There is no mod file that would simplify this.

when I run a simulation the program dies without giving an error.

Not very helpful. In such cases, it is up to the programmer to locate the cause of the problem. Did you happen to develop your program in an incremental manner, starting with something simple that works, then proceeding iteratively, that is,

Code: Select all

REPEAT
  make a small change
  test to make sure that the latest change works
UNTIL the whole program exists and works correctly

If not, then maybe you can embed print statements at key points in your program, then execute it to see where the failure occurs, and use that information to refine your bug search.

I think you mentioned this already in your comment:

If synapses are scattered over multiple nodes, you'll just have to keep track of them when you set them up, and compute a "sum on the fly" as the simulation advances.

Can you elaborate on that or send some code?

First, my opinion now is that I wouldn't bother adding up the currents during a run. I'd capture their time courses to hoc Vectors, and add those up after the simulation ends.

"How do I keep track of the synapses as I set them up?"

Append each synaptic mechanism to a list when you create it. Then you can iterate over the contents of that list without having to keep track of "magic numbers" such as the number of synaptic mechanisms you created. Example:

Code: Select all

# create the instances of the synaptic mechanism
syns = [] # syns is a Python list
NSYN = 3
for ii in range(0,NSYN):
  tmp = h.AlphaSynapse(0.5, sec = soma)
  tmp.onset = 1.0+ii*0.4
  tmp.gmax = 1e-3
  tmp.e = 0
  syns.append(tmp)

Now you can set up Vector record by doing this

Code: Select all

ivecs = [] # for the individual synaptic currents
for abc in syns:
  tmp = h.Vector()
  tmp.record(abc._ref_i)
  ivecs.append(tmp)

Notice that I'm iterating over the elements of one list in order to populate another list. Also notice the absence of magic numbers--not a 3 nor its symbolic representation NSYN in sight. No need to remember either the number or what you called it.

"OK, how do I add them up after the simulation ends?"

Code: Select all

def myrun():
  h.run() # initializes and runs a simulation, and also updates NEURON's InterViews graphs
    # you may prefer to call finitialize and continuerun directly
  itotal = np.array(ivecs[0]) # get the first Vector into a numpy array
  for ii in range(1,len(ivecs)):
    itotal += ivecs[ii].as_numpy()
  #
  pyplot.plot(tvec.as_numpy(), itotal)
  pyplot.show()

[ted]

After a bit of reflection, I think this version of myrun() is preferable. Uses no indices, and treats all members of the list in exactly the same way.

Code: Select all

def myrun():
  h.run()
  itotal = np.zeros(len(tvec))
  for abc in ivecs:
    itotal += abc.as_numpy()
  #
  pyplot.plot(tvec.as_numpy(), itotal)
  pyplot.show()

[ted]

Here are the hoc equivalents for the Python code in my previous posts.

"How do I keep track of the synapses as I set them up?"

Append each synaptic mechanism to a list when you create it.

Code: Select all

objref syns, tmp
syns = new List()
for ii=0,NSYN-1 {
  soma tmp = new AlphaSynapse(0.5)
  tmp.onset = 1.0+ii*0.4
  tmp.gmax = 1e-3
  tmp.e = 0
  syns.append(tmp)
}
objref tmp // so tmp can't be (mis)used to break anything

Now you can set up Vector record by doing this

Code: Select all

objref tvec, ivecs
tvec = new Vector()
tvec.record(&t)objref itotal, g

proc myrun() { local ii
  run()
  itotal = new Vector(tvec.size(),0)
  for ii = 0,ivecs.count()-1 itotal.add(ivecs.o(ii))
  g = new Graph()
  itotal.plot(g, tvec)
  g.exec_menu("View = plot")
}
ivecs = new List()

for ii = 0,syns.count()-1 {
  tmp = new Vector()
  tmp.record(&syns.o(ii).i)
  ivecs.append(tmp)
}
objref tmp

There's no way to avoid using an index variable to access individual items in a hoc List. Even so, it's better to use Lists rather than "arrays" of objrefs.

"OK, how do I add them up after the simulation ends?"

Code: Select all

objref itotal, g

proc myrun() { local ii
  run()
  itotal = new Vector(tvec.size(),0)
  for ii = 0,ivecs.count()-1 itotal.add(ivecs.o(ii))
  g = new Graph()
  itotal.plot(g, tvec)
  g.exec_menu("View = plot")
}