Impedance-class with point-processes?

[eacheon]

Since version 5.3, when the second argument is 1, an extended impedance calculation is performed which takes into account the effect of differential gating states. ie. the linearized cy' = f(y) system is used where y is all the membrane potentials plus all the states in KINETIC and DERIVATIVE blocks of membrane mechanisms.

Is "membrane mechanisms" in the last sentence meaning all mechanisms including point-processes in the model, say, a synaptic conductance (GABA synapses as point-process which have DERIVATIVE block for example), or all density mechanisms excluding point-processes?

[hines]

Here, membrane mechanisms refers to both density mechanisms and point processes. If there is any context in
the documentation where it is restricted
to density mechanisms, let me know
so I can disambiguate it. There may be some places where I include artificial cells as well and that would be a mistake. All three together should be called mod file mechanisms.

[eacheon]

Here, membrane mechanisms refers to both density mechanisms and point processes. If there is any context in
the documentation where it is restricted
to density mechanisms, let me know
so I can disambiguate it. There may be some places where I include artificial cells as well and that would be a mistake. All three together should be called mod file mechanisms.

Thanks. My confusing comes from my previous understanding to the phase "membrane mechanisms". Two more questions I have with regard the Impedance class:

1. I suppose either using .compute(varfreq, 1) or .compute(freq), NEURON is taken into account the pointprocess and density mechanisms. In other words, as the red text indicates, using .compute(freq) would take all mechanisms into consideration. Am I understanding the document wrong?

2. Then what is the difference between these two syntax the blue text is trying to convey? Is it that .compute(freq) uses membrane mechanisms, but does not use "states in KINETIC and DERIVATIVE blocks of membrane mechanisms"? I find this hard to understand.

DESCRIPTION
Transfer impedance between location specified above and any other location is computed. Also the input impedance at all locations is computed -- v(x)/i(x) Frequency specified in Hz. All membrane conductances are computed and used in the calculation as if fcurrent() was called. The compute call is expensive but as a rule of thumb is not as expensive as fadvance().

Since version 5.3, when the second argument is 1, an extended impedance calculation is performed which takes into account the effect of differential gating states. ie. the linearized cy' = f(y) system is used where y is all the membrane potentials plus all the states in KINETIC and DERIVATIVE blocks of membrane mechanisms. Currently, the system must be computable with the Cvode method, i.e.extracellular and LinearMechanism are not allowed. See deltafac

2. By try and error, I think document here has an flaw:

compute

Impedance

SYNTAX
.compute(freq)
.compute(&varfreq, 1)

[hines]

point 2 first. The freq arg is call by
value in both cases. You'll have to send
me an example where passing a reference did not give an immediate error.

Now point 1. I direct your attention to the longer discussion for the Impedance class itself:
http://www.neuron.yale.edu/neuron/stati ... edanc.html
In both cases one computes the conductance of both density mechanisms and point processes. Remember that in the first case, all states are held constant. In the second case and the small signal limit the conductances are changing sinusoidally at the same frequency as the voltage but with a different phase. Sometimes resulting in an apparent negative conductance.

[eacheon]

point 2 first. The freq arg is call by
value in both cases. You'll have to send
me an example where passing a reference did not give an immediate error.

I only have examples that gives immediate errors.

Now point 1. I direct your attention to the longer discussion for the Impedance class itself:
http://www.neuron.yale.edu/neuron/stati ... edanc.html
In both cases one computes the conductance of both density mechanisms and point processes. Remember that in the first case, all states are held constant. In the second case and the small signal limit the conductances are changing sinusoidally at the same frequency as the voltage but with a different phase. Sometimes resulting in an apparent negative conductance.

This makes it very explicit now. Thanks you for the explanation.

[christina]

Hello again Ted & Michael,

Rather than starting a new thread, this one seemed a reasonable place to ask this question.

As stated in the discussion above, NEURON is able to include the contributions of both active membrane channels and synaptic conductances into an impedance calculation. However, in the detailed Impedance documentation, under “Bugsâ€

[ted]

First, this really isn't a bug, it's an approximation. Second, this approximation
is made in the context of a whole host of other approximations such as:
1. Impedance estimates are only approximations to the behavior of a
nonlinear system over a narrow range of state variables in which, one hopes,
it is possible to describe the system by a linearized model.
2. The anatomical and biophysical properties of your model are themselves
only accurate to 5% at best (e.g. morphometric data gathered with light
microscopy seem to assume that all neurites have a circular cross-section,
but EM shows that most are at least ellipsoidal, and some are concave-
convex or highly irregular).

Third, you don't have to worry about effects of concentration changes on
membrane properties unless you expect significant concentration changes
to occur in the course of synaptic input, or to be induced by that synaptic
input. You can discover for yourself if this is likely to happen, if you have a
model that includes a ca accumulation mechanism. Just run a simulation
during which you activate your synapses of interest, and see whether cai
is affected, and if so, what it does to any ca-gated conductances. Even if
there is some effect, it will probably be (1) very small, and (2) dominated
by low frequency components, far below the range of frequencies that are
relevant to unitary synaptic events, therefore unlikely to affect peak psp
amplitude.

[christina]

Ted,

Yes, you are right, this is an *approximation* and not a bug. You might want to change the wording on the webpage to state that.

Your suggestions are helpful. I'll simulate some synapses explicitly to make sure the concentrations do not affect the membrane properties.

Thanks for the prompt reply.

-Christina

[ted]

Documentation of UNIX/Linux software often uses "bug" in a rather loose
sense. Laconic if not cryptic, it relies on readers to infer shades of meaning
and implications that are not explicitly stated. From that perspective, anything
short of "100% robust and approaching the theoretical limit of precision" is
often labeled a "bug" in lieu of a more detailed explanation. Some might
regard such documentation itself as buggy. But "bug" seems positively
refreshing compared to "issue," a weasel word that I suspect was originally
crafted by some focus group of mealymouthed liability lawyers.