update script for 2022 version of allen, remove allensdk

[PolarBean]

Description

Here we add the 2022 version of the Allen. We also remove the allensdk as it is becoming increasingly incompatible with other parts of brainglobe (getting it to work in Python3.13 while supporting threading was too hard)
What is this PR

• Bug fix
• Addition of a new feature
• Other

References

#233

Does this PR require an update to the documentation?

Yes we will update the atlas docs.

Checklist:

• The code has been tested locally
• Tests have been added to cover all new functionality (unit & integration)
• The documentation has been updated to reflect any changes
• The code has been formatted with pre-commit

[PolarBean]

25 micron file
removed as ill provide a version with smoothed meshes

[PolarBean]

works at 25um. I'll post the 10um when its done.

[PolarBean]

The allen have done a really fantastic job making smooth meshes. I cant really get comparable quality. I'll just reuse their meshes and restrict the mesh generation to regions that have been updated. Does anyone know if they have open sourced their mesh generation pipeline as I think it could be really useful, for other atlases also.

[adamltyson]

IIRC their mesh generation isn't fully automatic. The reason the meshes are so nice is because lots of work went into smoothing them manually.

[PolarBean]

25um: https://data-proxy.ebrains.eu/api/permalinks/b68c5095-2779-4240-9ded-a3ce91609b4b
10um: https://data-proxy.ebrains.eu/api/permalinks/801d1bb3-239f-49e3-972e-b8cf73063d2d
both pass validation

[PolarBean]

both work in napari

[PolarBean]

Hmm so the create_region_mesh function expects a tuple instead of normal arguments (this is for multiprocessing purposes). This is ok but it makes it hard to add new optional arguments to it. Could we instead make a wrapper function which takes a tuple and calls the function which now has normal arguments?

[adamltyson]

Hey @PolarBean, probably best to create a new issue and we can discuss. The mesh generation functions are long overdue some more thought. @brainglobe is on holiday for two weeks now though, so we'll take a look at this PR in Jan.

[PolarBean]

ah ok. I'll make a new issue. I modified it pretty minimally so that I could add the new argument whilst keeping it backwards compatible and keep all of the tests passing. No rush at all in merging this, happy christmas

[adamltyson]

Hey @PolarBean, just getting back to this. A couple of questions:

• I assume this works at 50um too?
• Is it possible to address [Feature] refactor mesh utilities #729 first? Then we can simplify this PR a bit.

[PolarBean]

yeah ill generate 50 and 100, good catch. Also yes, I'll make a separate pr for the mesh generation changes.