52 control vocabulary for targeted structures

[dbirman]

Adds a new Mouse CCF class, use by doing:

from aind_data_schema_models.mouse_ccf import CCFStructure

and then:

CCFStructure.ROOT
CCFStructure.VISP

etc

Also available are:

CCFStructure.from_id(997)
CCFStructure.from_acronym('root')
CCFStructure.acronym2id("root")
CCFStructure.id2acronym(997)

[dyf]

I'm a little nervous calling the CCF terms a registry, given that we historically called it an "ontology" and now some are pushing for "parcellation terminology".

Suspect we might have to change this later, but likely we will want to wholesale rename Registry to something else, so I guess we can proceed as-is.

[saskiad]

I'm a little nervous calling the CCF terms a registry, given that we historically called it an "ontology" and now some are pushing for "parcellation terminology".

Suspect we might have to change this later, but likely we will want to wholesale rename Registry to something else, so I guess we can proceed as-is.

Is your concern about having it in a list of registries? Or in having the field that references it be registry? Because the latter is vestigial from using PIDName and I'm thinking we don't need to be super strict on that. We can call the field "ontology" in its place. (I'm not sure parcellation terminology is going to work well).

[saskiad]

Okay, @dyf and I hashed this out

• let's not add CCF to the registries
• in the CCFStructureModel have these fields:
• - atlas (hard coded to CCF v3)
• - name
• - acronym
• - id

Future issue will be to get rid of the concept of a Registry and adjust the schema to point to taxonomies more directly. Since we aren't using the PIDName structure we can do this more easily. This will be a big change, but I think it will be better (and reduce nesting)

[dbirman]

Okay those changes are in place, either of you can merge this and maybe make a ticket to track the larger change if you didn't already?

[dbirman]

After discussion w/ Saskia we decided:

1. Not to change the base CSV files, they just have floats for now (these are pulled directly from BG)
2. To refactor the model to a single class that can be shared across different atlases

See brain_atlas.py for how I built CCFStructure, other atlases can easily be added by copying the CSV file and repeating that pattern. It would be very quick to add all the remaining BG atlases now (rat, etc).

To do this I also had to add a new feature to create_literal_class that lets you pass a value that is shared.