Validate atlas name

[adamltyson]

See #523 for context

To do

• add more stringent atlas name validation to validate_atlas_name in brainglobe_atlasapi\atlas_generation\validate_atlases.py.
• adjust repr method which currently allows more flexible atlas names than what should be allowed.
• add test with new constraints to tests\atlasgen\test_validation.py

[stellaprins]

The following atlas names are already covered :

• nadkarni_mri_mouselemur_91um
• example_mouse_100um
• axolotl_50um

resulting in

• nadkarni mri mouselemur atlas (res. 91um)
• example mouse atlas (res. 100um)
• axolotl atlas (res. 50um)

When I fixed the bug in the str method allowed atlas names that do not end with the resolution so I wanted to add a test that checks whether 50um_axolotl results in 50um axolotl atlas.

However after talking to @alessandrofelder and checking the atlas names on last_versions.conf (GIN) -- they all end with the resolution -- we decided it could make more sense to always expect atlas names to end with the resolution and validate this (if this is not already happening).

Just wanted to check what you're opinion @adamltyson and whether you can think of any other (edge) cases that should be added to the test.

[adamltyson]

However after talking to @alessandrofelder and checking the atlas names on last_versions.conf (GIN) -- they all end with the resolution -- we decided it could make more sense to always expect atlas names to end with the resolution and validate this (if this is not already happening).

Agree. The whole point of the Atlas API is standardisation, so we should validate/enforce it where we can.

[stellaprins]

Atlas naming convention

allowed name structures

• [source_id]_[species]_[resolution]
• [source_id]_[species]_[additional_info]_[resolution]

allowed characters

• Lowercase letters (a-z)
• Uppercase letters (A-Z) (see note 2)
• Underscores (_), hyphens (-) and periods (.)

Elements in atlas name

• [source_id] e.g., allen, kim, osten
• [species] e.g., mouse, zfish, human
• [additional_info] some atlas names include details about the developmental stage and / or specific imaging technique used (e.g., e11-5, p14_mri-MTR, lsfm)
• [resolution] one or more digits, optionally followed by a period and one or more digits, and ending with "um". (e.g., 37.5um, 100um)

Deduced from looking at the atlas names here https://gin.g-node.org/BrainGlobe/atlases/src/master/last_versions.conf

Note1: @alessandrofelder told me that atlasses with a "mm" resolution will probably be added in the future.

Note 2: kim_dev_mouse_p14_mri-MTR_50um is an atlas I found on GIN last_versions.conf with an invalid name (it contains uppercase letters)

[stellaprins]

Just found out there already is a validate_atlas_name function which is validates that atlas names do not contain upper case letters. Will add the additional constraints I summarised above under naming convententions.

def validate_atlas_name(atlas: BrainGlobeAtlas):
    """
    Ensures atlas names are all lowercase
    """
    assert (
        atlas.atlas_name == atlas.atlas_name.lower()
    ), f"Atlas name {atlas.atlas_name} cannot contain capitals."
    return True