Distance function does not calculate path length

[bantin]

Hi @michaeldeistler and jaxley team -- I'm working with a few folks to port the biophysical model implemented in this paper from NEURON to Jaxley.

Crucial to this model is that the channel conductances vary by distance to the Soma, where distance is the path length along the tree. Looking at the source code for base.py it seems like the distance returned by Jaxley is just the euclidean distance. Is this true? If so, I'd like to help remedy this!

A second and related question, is that it's not very clear to me what the right way is to iterate over a cell and calculate the distance from each compartment to the soma. Something that I'd expect to work is roughly:

 for branch in cell.branches:
        for c in branch.compartments:
            # get distance to soma

but this doesn't work. Would you mind clarifying what the right way to do this is?

Thanks a bunch! Ben

[michaeldeistler]

Hi Ben!

Thanks for reaching out, and great that you are starting to look into Jaxley :)

Is this true? If so, I'd like to help remedy this!

Indeed, the distance method computes the euclidean distance. Path distance is not implemented yet. @jnsbck will merging #355 automatically allow to compute the distance along a path?

not very clear to me what the right way is to iterate over a cell and calculate the distance from each compartment to the soma

Your code almost works: it's comps instead of compartments:

import jaxley as jx

comp = jx.Compartment()
branch = jx.Branch(comp, ncomp=4)
cell = jx.Cell(branch, parents=[-1, 0, 0])

for branch in cell.branches:
    for comp in branch.comps:
        comp.set("radius", 2.0)

or

for branch in cell:
    for comp in branch:
        comp.set("length", 200.0)

This requires the most recent version of Jaxley (v0.5.0). Also see this tutorial.

Hope this helps! Let me know if anything is unclear!
Michael

[bantin]

Awesome, thanks @michaeldeistler !

Hoping #355 is merged soon. In the meantime, I may check out that branch and try getting things working there.

[jnsbck]

Hey Ben,
yes indeed merging #355 would make this trivial, which will hopefully happen next week if @michaeldeistler finds time to review the PR (fingers crossed). @manuelgloeckler even implemented a draft for path distances a while back here.

Jonas

[jnsbck]

#355 is merged now. Have a look here for how to use it :)

[chaseking]

Thank you both! What's your PyPI release schedule? Using a local, git pulled jaxley isn't a huge burden but just curious

[michaeldeistler]

We have a dev meeting scheduled Wednesday and I think we can release after that.

Michael

[michaeldeistler]

New version is on PyPI now. Please let us know if anything does not work as expected!