NEURON scripting exercises
Exercise 1
Given a list of sections:
from neuron import h, gui
secs = [
h.Section(name='soma'),
h.Section(name='dend'),
h.Section(name='axon')
]
and given some data
data = {
"axon": {"L": 100, "diam": 1},
"soma": {"L": 10, "diam": 10},
"dend": {"L": 50, "diam": 3}
}
Write a script that uses the data to set the L
and diam
parameters for each Section. Set the number of segments such that there is (1) an odd number of segments and (2) each segment is as close to 5 microns long as possible without going over. Connect all sections not named soma
to the center of the soma.
Use an PlotShape
to check your work. Store the object as, e.g. ps
. Then use ps.show(0)
to display the diameters.
Bonus challenge: Move the data into a JSON file and use Python’s json module to read it from the file.
Exercise 2
Simulate and visualize a propagating action potential on a long cable with Hodgkin-Huxley dynamics triggered by a current clamp.
Things to consider:
What is an appropriate diameter, length of cable, and current injection amplitude?
How can you best illustrate a propagating action potential in a single figure?
What discretization should you use?
Exercise 3
Continuing with the previous exercise, record and plot the membrane potential, sodium current, and potassium current at the center of the cable as a function of time.
Hint: Use the Vector
class’ record
method.
Things to consider:
How does the relationship between the sodium and potassium currents differ in a Hodgkin-Huxley simulation from the relationship in mammalian cells?
Exercise 4
Construct two Y-shaped cells, where each branch is 2 µm in diameter and 100 µm long. Add Hodgkin-Huxley channels with the default parameters. Inject current into one cell and ensure that you are able to generate a propagating action potential in that cell while the other one remains at rest.
Connect the two cells with a gap junction (see halfgap.mod) at a location of your choosing. A 3 MΩ resistance as in halfgap.mod
should allow an action potential in one cell to initiate an action potential in the other. Verify this. What is the strongest resistance that still allows action potential initiation in the second cell?
Bonus challenge: Make each cell an instance of a Python class.
Bonus challenge: Modify the class to allow repositioning (moving and/or rotating) the cells.
Exercise 5
Download the model from https://modeldb.science/126814 Create a Python script that loads the model, injects current into the center of the soma from t=2 to t=4 ms sufficient to generate an action potential, and records and plots membrane potential, sodium current, and potassium current as functions of time from t=0 to t=10 ms.
Hint: h.load_file('mosinit.hoc')
Hint: This model uses the variable time step solver. Be sure to record t
or switch to a fixed step solver with h.cvode_active(False)
.
Bonus challenge: Export the recorded data to a CSV file and then open and plot it in Excel, MATLAB, or a similar tool of your choice.
Bonus challenge: How does the plotted data change as h.celsius
is varied?
Exercise 6
NEURON’s ExpSyn
mechanism generates synaptic currents of the form
where
What is the role of \(E_syn\)? How does it change for an excitatory vs an inhibitory synapse? (Note, this is the parameter e
in the code.
Construct two single compartment neurons with Hodgkin-Huxley dynamics, one of which receives a current pulse at 2 ms, another which receives a current pulse at 10 ms. Ensure that both cells fire action potentials after the input.
Now, using NetCon
and ExpSyn
, construct an inhibitory synapse between the two with the cell that fires later as the post-synaptic cell. Choose a delay and strength such that the post-synaptic cell is inhibited from firing. Plot the membrane potentials vs time.
halfgap.mod
NEURON {
POINT_PROCESS HalfGap
POINTER vgap
RANGE r, i
ELECTRODE_CURRENT i
}
PARAMETER {r = 3 (megohm)}
ASSIGNED {
v (millivolt)
vgap (millivolt)
i (nanoamp)
}
BREAKPOINT { i = (vgap - v) / r }
Note: we will see another way to generate gap junctions using the Linear Circuit Builder later in the course that is more numerically stable than the approach shown here.