Problem in importing 3D data to Neuron

[Sherif]

Hi,

I am trying to import a 3D reconstructed morphology (.asc file) of a neuron using the 3D import tool in Neuron. After importing the file I got a message from Neuron on possible roon-branch errors in which it suggests connecting some branches to nearest soma. I looked at the values of these offset distances and found all of them to be less than 5 um, thus I was comfortable accepting to connect these branches to the soma. However, when I checked the boxes I got error message for each checked box. The error message says:

nrniv: Matrix index out of range: index=31 max_index=30
near line 2
{Import3d_GUI[0].redraw()}
^
Import3d_Section[0].pl_stk_triang(Graph[0]Import3d_Section[0]... , , )
Import3d_Section[0].pl_diam(Graph[0] )
Import3d_GUI[0].pl_diam( )
Import3d_GUI[0].pl( )
and others
initcode failed with 3 left

My question is why Neuron is complaining about these connections?

When I tried to import the 3D morphology to the cellbuilder, the cell was reduced to a single straight line. Surprisigly, when I unchecked the "Show Points" and "Show Diam" boxes, a 3D morphology appeared. Does this morphology have the suggested connections between the branches with offsets and the soma, I don't know.

Does anybody have an explanation to these events/messages?

Thank you very much.

[ted]

That's very strange stuff, Sherif. I'll have to see that for myself--can you zip up the .asc file and send it to me? Also, which version of NEURON are you using?

[hines]

The file seems to have many duplicate, partial, and out of order (non-monotonic in z axis)
contours that define the soma. I have not added any contour manipulation to the import3d tool so you will have
to manually assess and get rid of the ones that should not be there and properly order the
rest. It does not appear to me that you should click any of the "Possible root branch errors" checkboxes
since all the dendrites should be connected to the center of the soma and NOT to the contours themselves.
Apparently the import3d tool was tricked into thinking a contour was equivalent to a dendrite and I will
eventually have to fix that. I guess the last functionality enhancement (contour stacks) needs a bit of
work. The "Possible root branch error" means that the branch is not in the soma bounding box and is
also not extremely close to another branch end. In this case, the default is to still connect to the soma
but offer the possibility of connecting to the nearest branch.
At present I have no idea how to deal with all the contours in this file. I guess my first step would be
to extend the import3d tool so they can be selected and thrown away.

Can you tell me what the distinctions are between
("coutour coupe S"
("canal central S"
("CellBody"
("contour soma2"
etc.
Import3d throws away all contours except "CellBody"

There are 5 contours with the label
("CellBody"
and they have z-axis values of
-0.00
-5.87
-10.77
3.23
6.85 (but with a first point at 7.83)
putting the last first (and changing the first point to 6.85) and the second-to-last second
makes for a very nice looking cell body.