Biology Modeling FAQ

How do I work with neuron morphologies?

NeuroMorpho.Org hosts over 170k reconstructed neurons, all of which are available in SWC format as well as their original format.

NEURON’s Import3D tool can read SWC files as well as several other neuron morphology types.

The Import3D tool can be used both through the GUI and programmatically, however it is generally best to start with the GUI to explore the morphologies and make sure they are suitable for use with simulation. See the Import3D GUI tutorial for more on GUI usage.

Example:

To create pyr, a Pyramidal cell object with morphology from a file called c91662.swc:

from neuron import h
h.load_file("stdlib.hoc")
h.load_file("import3d.hoc")

class Pyramidal:
    def __init__(self):
        self.load_morphology()
        # do discretization, ion channels, etc
    def load_morphology(self):
        cell = h.Import3d_SWC_read()
        cell.input("c91662.swc")
        i3d = h.Import3d_GUI(cell, False)
        i3d.instantiate(self)

pyr = Pyramidal()

Here pyr has lists of Section objects: pyr.apic, pyr.axon, pyr.soma, pyr.all. Each section has the appropriate .name() and .cell().

Note: this example is for an SWC file specifically; other readers are supported for different formats including h.Import3d_Neurolucida3(), h.Import3d_MorphML(), and h.Import3d_Eutectic_read().

Note: if multiple cells are instantiated from the same reconstruction, they will occupy the same spatial locations unless they are explicitly translated.

How do I simulate a current clamp pulse experiment?

Use an IClamp at a segment of your choice. You can specify the delay (in ms) (when the clamp starts), the amp (amplitude in nA) of the current pulse and the dur (duration in ms). The injected current may be monitored through the IClamp’s i state, or recorded using its _ref_i.

For example, the following code specifies that a current of 10 nA will be injected directly into the center of the soma from t = 1 ms to t = 1.1 ms:

from neuron import h
from neuron.units import ms

# setup the model here

ic = h.IClamp(soma(0.5))
ic.amp = 10
ic.dur = 0.1 * ms
ic.delay = 1 * ms

Note

You must keep the ic object accessible in some way (either by assigning it to a variable or storing it in a list); if it becomes inaccessible, then Python will “garbage collect” it and it will be removed from the simulation.

For a runnable example, see these tutorial exercises.

How do I simulate a current clamp with non-pulse behavior?

In this scenario, you likely have or can construct two pairs of Vector objects: i_stim with the injected current (in nA) measured at various time points and t_stim the corresponding time points (in ms).

Note

If instead of Vectors, you have Python lists, numpy arrays, or other iterables, you can get an equivalent Vector via

t_stim_vec = h.Vector(t_stim)

Use an h. IClamp at a segment of your choice as described above, set the delay (start time) to 0, the dur ation to a large number (e.g. 1e9) and use Vector.play() to play into the _ref_amp field using interpolation (the True in the following); e.g.

ic = h.IClamp(soma(0.5))
ic.delay = 0
ic.dur = 1e9
i_stim.play(ic._ref_amp, t_stim, True)

For a runnable example, see this example.

How do I make cytosolic concentrations change in response to ion channel activity?

NEURON defaults to assuming homeostatic mechanisms maintain intracellular concentration as that is often the assumption made by modelers, however this can easily be changed when the circumstances warrant.

For certain ions (e.g. calcium) the changes due to channel activity are significant. Likewise, in pathological conditions (e.g. ischemic stroke), even ions like sodium and potassium may show significant change.

To specify that intracellular sodium concentration on all sections (allsec()) is to be affected by ion channel activity:

from neuron import rxd
cyt = rxd.Region(h.allsec(), name="cyt", nrn_region="i")
na = rxd.Species(cyt, name="na", charge=1)

Here the nrn_region="i" indicates that we are talking about the intracellular concentration. The name argument to rxd.Species specifies the name of the ion. By default rxd.Region assumes that we’re describing a Region filling the entire Section; but this can be altered with additional arguments. The charge=1 corresponds to the fact that sodium ions have a charge of +1. By contrast calcium ions have a charge of +2, and thus to tell NEURON to consider how calcium changes due to ion channel activity we write:

ca = rxd.Species(cyt, name="ca", atolscale=1e-6, charge=2)

Here we have also added the optional parameter atolscale. It has no effect in fixed-step simulations, but for variable step simulations (see CVode) it is a hint that concentrations for calcium are often much smaller than those for sodium and that it should seek much smaller errors in calcium in terms of absolute numbers.

(As an aside, it is generally good practice not to use allsec() but to instead explicitly identify the sections to be used. NEURON provides the Section.wholetree() method for getting a Python list of all sections that belong to a cell containing a specified section. It would be natural to include specification that concentration is to change on a per-cell basis within a cell class; this compartmentalization allows combining cells from different models where we may want to make different assumptions.)

How do I make cytosolic concentrations diffuse and respond to ion channel activity?

We modify the above example by specifying a diffusion constant d e.g.

from neuron.units import um, ms
ca = rxd.Species(cyt, name="ca", d=1.3 * um**2/ms, charge=2)

The units used here – µm 2 / ms – are the default and would be assumed if not specified, but it is generally good practice to include units. We note that the neuron.units module provides both µm and um; these are synonyms with the latter made available to facilitate typing.