Reconstruction cookbook?

[vladimirov]

Dear colleagues,
Does anybody know a good manual for reconstruction of anatomically detailed models? I am new in the field, and there are some immediate questions to the topic:
1) After I open the swc file in import3D and save it as Neuron hoc file, is it good enough to populate with ion channels and do simulations?
2) I guess, answer to Q.1 is NO in many cases. What are the necessary checks to do with the tree?
a) I know that soma must be non-zero length (happens in some swc files)
b) are 3-ple or 4-ple forks of processes allowed? Or only simple forks (--<) ? Any requirements from the cable equation solver?
c) check for loops?
d) ...?
Please, advice what are other common "quality-control" steps for making a good Neuron tree out of raw SWC file.
Thank you!

[ted]

Excellent questions. I'm glad you are thinking about these issues.

1) After I open the swc file in import3D and save it as Neuron hoc file, is it good enough to populate with ion channels and do simulations?

The general answer is "maybe, maybe not."

Do you have any practical experience in microscopic neuroanatomy? If not, it would be a very good idea to find a colleague at your institution who does, and ask that person how it is done, what are criteria for "quality," and what caveats apply--maybe even offer to do a reconstruction for the sake of gaining experience. You might go to MicroBrightfield's web site and read whatever they post there about NeuroLucida and how to use it. Also, there is a scientific literature about the quality of morphometric data from the perspective of computational modeling; the two articles cited in this post
viewtopic.php?f=13&t=405&p=1215#p1215
could serve as a starting point
You might also want to look at Chapter 6. Methods in Corinne Teeter's dissertation
Characterizing the Spatial Density Functions of Neural Arbors
which is available as
http://gradworks.umi.com/3432040.pdf

With regard to your questions that are "NEURON-specific"--

2) What are the necessary checks to do with the tree?

At a minimum: discontinuities, zero diameter points, minimum diameter measured (some anatomists address questions for which diameter is unimportant, so their files may or may not have accurate diameter measurements). But also be sure to read the papers mentioned above.

a) I know that soma must be non-zero length (happens in some swc files)

Easily fixed by manual editing.

b) are 3-ple or 4-ple forks of processes allowed? Or only simple forks (--<) ? Any requirements from the cable equation solver?

No restriction on number of sections that may be connected at a single node.

c) check for loops?

NEURON does that and will complain if any exist.