[HackathonMar21] Test and improve Arbor<->NeuroML interactions

[pgleeson]

See: https://arbor.readthedocs.io/en/latest/fileformat/neuroml.html
https://github.com/OpenSourceBrain/ArborShowcase
https://github.com/thorstenhater/arbor-gui

[pgleeson]

@thorstenhater I've installed arbor-gui and managed to load a few nml files, including a newly generated example here: https://github.com/OpenSourceBrain/ArborShowcase/blob/main/NeuroML2/BallAndStick.cell.nml.

The points/morphology is getting imported fine, but the segment groups are not correct, e.g. the soma is going in the dendrite_group too, etc.

[thorstenhater]

Good to hear you got it working, if you have any comments I can add to the Readme to ease the process for the next person,
please tell me.

Thanks for catching that, although I assume that our NML import is at fault here, rather than the display. I will open an issue against arbor.

[thorstenhater]

OK, I got down to the bottom of this with the help of @halfflat: Turning a morphology into line segments sometimes returns partial segments. Currently the renderer does not deal with that and shows the full segment, even if the to be shown fraction is zero.

[pgleeson]

Good progress after chatting with @noraabiakar about cell compartmentalisation.
Aell in NML originally was simulated in Neuron (left) and Arbor (right) and produced the same output:

[thorstenhater]

Hi, the rendering bug has been partially addressed. It remains to be seen whether the more complex case needs to be addressed.