Model Control: Simulation Families

Modeling projects often involve executing a series of simulation runs, analyzing the output of each run, and summarizing the results. This exercise will have you examine how the location of an excitatory synapse affects the peak depolarization at the soma. In doing this, you will learn:

Model

This exercise uses the ball and stick model you have already seen.

Simulation

In the course/family directory, start NEURON with the init.hoc argument (UNIX: nrngui init.hoc , MSWin: double click on init.hoc). This particular init file
// start the GUI and load the ballstk cellbuilder
load_file("nrngui.hoc")
load_file("ballstk_builder.ses")

// load an initial starting session
load_file("start.ses")
makes NEURON read the session file for a CellBuilder that contains the specification of the model. You will use this CellBuilder to adjust the model parameters.
Note: Make sure the CellBuilder's Continuous Create button is checked. Otherwise the sections of the ball and stick model will not exist.

Using the CellBuilder to manage the spatial grid

1. Geometry/Specify Strategy : select d_lambda for the all subset. Also make sure that no section has an overriding strategy.
2. toggle Specify Strategy off
3. make sure that d_lambda = 0.1 space constant at 100 Hz

How many segments were created?
forall print secname(), " nseg = ", nseg
Where are the nodes located?
dend for (x) print x, " ", x*L

If these locations aren't particularly "esthetic," you can assign nseg a new (larger) value manually (odd multiples of 5 are nice). You could do this with a hoc statement like
dend for (x) print x, " ", x*L
but this would be a potential cause of confusion, so you should specify nseg through the CellBuilder instead.

Remember, you are using the CellBuilder with Continuous Create on. This means that, if you change the model specification in the CellBuilder, or even just toggle Continuous Create off and on, what you did in hoc code could be overridden.

Using the Representation

1. BUILD THE GUI

Set up a graphical interface that lets you apply an AlphaSynapse to the model (time constant 1.0 ms, peak conductance 0.005 umho) while observing Vm at the middle of the soma.

2. TEST THE MODEL

Put the synapse at the proximal end of the dendrite, turn on Keep Lines in the voltage graph, and run a simulation. Then reduce the peak synaptic conductance to 0 and run another. Use View = plot to get a better look at somatic Vm.

What's wrong with these responses? (hint: increase Tstop to 50 ms and run another simulation)
Change dendritic e_pas to -65 mV (use the CellBuilder's Biophysics page!) and try another run. Does this help? Why?

3. INITIAL EXPLORATION OF THE MODEL

Place the synapse at several different positions along the dendrite. Find and plot the peak amplitude of the somatic EPSP vs. synaptic location.

You will need the following:

When you are ready to use Gather Values, turn off Keep Lines in the graph of Vm vs. t. Then:
1. invoke putsyn(0)
2. L click on the voltage trace and move the crosshairs to the peak of the EPSP
3. L click on the Record Val button of the Gather Values tool

Repeat these 3 steps a couple of times, specifying a different location each time. After the second iteration you can use View = plot in the Gather Values graph to see how things are shaping up. Don't worry about the sequence of locations; you can always sort them in ascending order by L clicking on the "SortX" button.

4. SWITCHING TO PRODUCTION MODE

Manual data collection may be fine for initial exploration, but it's no way to pursue an extensive project that involves multiple experimental conditions (e.g. channel blockers) or complex cellular morphologies.

Here's an outline of one approach to automating the task:

For each node along dend
move the synapse to that node
run a simulation
determine the most depolarized somatic Vm for that location
save this value and the location in a pair of vectors
Plot the vector of peak values vs. the vector of locations
Use a text editor to create a procedure that implements this approach. Put it in a file called myprofile.hoc, and then use the command
xopen("myprofile.hoc")
or just use the NEURON Main Menu's "File / load hoc" button to make it available to your model.

You already have putsyn(), which takes care of the second and third items in this outline.
It may be helpful to know about :
for (x)    objref    new    the Vector class in general    Vector record()    Vector max()    Vector append()    Vector plot()    the Graph class

Here is a skeleton of one possible implementation of such a procedure.

5. THINGS TO TRY

1. Compute the actual EPSP amplitudes by subtracting the -65 mV baseline from the soma responses and plot the results.

2. Plot peak EPSP amplitude as a function of anatomical distance along dend in microns.

2. What would happen if the somatic HH currents were blocked? Use the CellBuilder to reduce gnabar_hh and gkbar_hh to 0. Make sure to change eleak to -65 mV before running a new series of simulations (why or why not? and what if you don't?).

Compare these results with what you saw when HH currents were not blocked. Do spike currents in the soma enhance all EPSPs, or does the nature of the effect depend on synaptic location?


NEURON hands-on course
Copyright © 1998-2001 by N.T. Carnevale and M.L. Hines, all rights reserved.