Page 1 of 1

import of neurolucida .asc

Posted: Wed Jul 27, 2005 1:35 pm
by ubartsch
Hi,
I have imported Neurolucida .asc files to Neuron either through Neuron itself or cvapp. Spines are omitted during import.
But the centroid view shows a very zig-zag like structure of the dendrite (based on the pt3dadd coordinates, I guess).

Why is the dendrite that zig-zag like if the spines are omitted? Shouldn't it be rather straight, like the "backbone" of the dendrite?!

Is there any difference in modeling between a similar cylindrical structure made from scratch (as seen in the schematic view of the imported morphology) and the imported structure?

Many thanks for a short answer!
Regards
UB

Re: import of neurolucida .asc

Posted: Wed Jul 27, 2005 2:55 pm
by ted
Why is the dendrite that zig-zag like if the spines are omitted? Shouldn't it be rather straight, like the "backbone" of the dendrite?!
Look at it in cvapp or NEURON's Import3D tool; if you don't see zigzags,
the problem is in cvapp or the Import3D tool. If so, please let me know.

But if you do see zigzags, you're probably dealing with corkscrew artifact,
which results from tissue shrinkage during fixation.

There are articles about quantitative morphometry of neurons that discuss
the various artifacts that can afflict such data. These are well worth reading.
Two highly recommended ones are
Kaspirzhny AV, Gogan P, Horcholle-Bossavit G, Tyc-Dumont S. 2002.
Neuronal morphology data bases: morphological noise and assesment
of data quality. Network: Computation in Neural Systems 13:357�380.
Scorcioni, R., Lazarewicz, M.T., and Ascoli, G.A. Quantitative morphometry
of hippocampal pyramidal cells: differences between anatomical classes and
reconstructing laboratories. Journal of Comparative Neurology 473:177-193, 2004.

You may never want to use morphometric data again. For that matter,
how good are the published biophysical characterizations of voltage- and
ligand-gated currents? (maybe that belongs in a separate discussion thread)

An observation:
People tend to be acutely aware of all the strengths and weaknesses, pros and
cons, gotchas and caveats, that pertain to the area in which they have expertise.
No experimental observation, logical inference, or theoretical prediction escapes
the minutest scrutiny and critical deconstruction. However, let them step just outside
this familiar territory, and the natural tendency is to accept anything they hear or see
as solid fact.
Is there any difference in modeling between a similar cylindrical structure made from scratch (as seen in the schematic view of the imported morphology) and the imported structure?
What does "schematic view of the imported morphology" mean?
The simplified figure displayed in the CellBuilder, where every section is
represented by a single straight line?

I think you already know the answer to your question: keep surface area,
diameter, and length unchanged, and the results will be the same.

Posted: Wed Jul 27, 2005 6:04 pm
by ubartsch
The zigzags do appear in cvapp and in Neuron,
- but they are not seen under Neuroexplorer from Neurolucida
- and they are not seen on the axon of the cell which has no spines!

So only spiny part s of the cell show that zigzag like shape.

To put this better I add some figures.
The cell in the Neuroexplorer tool:

Image

The same cell in cvapp:

Image


The dendrite looks like connection of spines one after another...?
If this due to the shrinkage this should be seen in the Axon as well (blue in the top fig, grey in the lower fig)?!

Posted: Wed Jul 27, 2005 8:11 pm
by ted
Ugly as a box of frogs. Can't say I've ever seen the like. The data are in
your hands, so you're best situated to discover what this is all about. That
is, pick a branch and try to make sense out of it. Does each xyz datum
signify the head of a spine? or is it the location along the dendrite where
the spine is attached? If the latter, those dendrites are fat. It would also
be a good idea to talk about this with the person who made the
measurements, or get an anatomist to look at it.

Posted: Thu Jul 28, 2005 7:15 am
by ubartsch
Spines seem to be coded by the position of their head. An example from my Neurolucida ascii file:

Code: Select all

( (Color Red)
  (Dendrite)
...
    (    8.77   -18.01   -11.75     0.87 S2)  ; R-1, 6
        <(    9.68   -16.96   -11.50     0.65 S2)>  ; Spine
...
   
No value equals another, so I guess the base is not coded.
This can be easily seen in another view in the Neuroexplorer tool where every coordinate is shown in the screen as a white square.

If I exclude some lines with "...; Spine" at the end , spines disappear in Neuroexplorer and the respective zigzag "caused" by the spine in the cvapp imported file as well!

Compare this figure to the figure from Neuroexplorer above:

Image


So, to be honest, I think there is a bug in the import function (borrowed from cvapp?). It does not omit spines but includes in to the dendrite in a wrong way, by taking spine coordinates as coordinates of the dendrite!
At least in case of my special file. Did nobody ever report a similar thing after importing Neurolucida files containing spiny dendrites? Is the version on Neurolucida important?

I guess the easiest way for me (and maybe for others) is to exclude all the spines ( delete all lines with <..>; Spine ), import the general structure and add the missed membrane area (which can be extracted through the Neuroexplorer Branch Analysis) to the imported structure through increasing the diam of the compartments in Neuron.

Using sed to eliminate troublesome lines from a file

Posted: Thu Jul 28, 2005 10:16 am
by ted
Thanks for digging into this problem. Looks like a bug that needs to be
fixed.

In the meantime, it's trivial for any user of a real OS (UNIX, Linux, OS X)
to filter out all lines that contain a particular string--awk and sed can do
this nicely, e.g.
sed -n '/Spine/!p' fil.dat > fixed.dat
will produce a file called fixed.dat that contains only those lines of fil.dat
which do not have "Spine".

MSWin users who have recent distributions of NEURON can do the same
thing, because installing NEURON also installs a subset of cygwin that
includes sed. First start a cygwin terminal by double clicking on rxvt.sh
in the NEURON program group. Then cd to the directory that contains
the flie that needs changing (note that the root of the primary hard drive,
which MSWin calls C:\, is /cygdrive/c). Finally enter the command string
(see above). To exit the shell, type ^D at the $ prompt.

Posted: Thu Jul 28, 2005 1:11 pm
by hines
Edit the nrn/lib/hoc/import3d/read_nlcda3.hoc file and search for
proc spine()
then change the "properties() points()" line so that it looks like

Code: Select all

//      properties() points()
        while (current.token != rightsp) {
                read_next_token()
        }
that will ignore the spines.
The import3d tool in the alpha version has some additional bug fixes (ignore marks) and rudimentary editing for z-axis problems. See
http://www.neuron.yale.edu/ftp/neuron/versions/alpha/