Design doc: Add anatomical ontologies documentation for DANDI#2768
Design doc: Add anatomical ontologies documentation for DANDI#2768bendichter wants to merge 6 commits into
Conversation
Document the strategy for anatomical ontologies in DANDI, outlining goals, available systems, and implementation phases for dandi-cli and DANDI Atlas Explorer.
|
|
@satra, I did some research on ontologies on Friday, trying to assess the capabilities of HOMBA against other available systems and come up with a strategy for integrating them. This design doc walks through the five relevant systems (Allen CCFv3, UBERON, HOMBA, CHARM, and HBA) — what each one covers, where each falls short — and proposes a layered, species-specific strategy that gives us consistent naming, cross-species queries, and spatial data for the Atlas Explorer. Would love to hear your thoughts. |
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
Could you also include Paxinos-Watson for rats? I'm working with someone at Dartmouth doing this now: https://neurosift.app/nwb?url=https://api.dandiarchive.org/api/assets/357f2323-6ed5-43ab-9559-fe454acbe7cd/download/&dandisetId=001765&dandisetVersion=draft&tab=/general/extracellular_ephys/electrodes (will have BEP032 version when that is possible as well) |
|
@CodyCBakerPhD yes, good call. Rats are missing from HOMBA, so I didn't consider them here, but you are right that we should include them as they are quite prominent in the archive |
|
|
||
| ## Landscape of available systems | ||
|
|
||
|  |
There was a problem hiding this comment.
was .png the original form produced or may be there is some .svg or even some exclalidraw or mermaid? it really feels we need vector graphic for it, at least .svg so it could be tuned later
|
|
||
| ### Allen Mouse Brain CCFv3 | ||
|
|
||
| The Allen Mouse Brain Common Coordinate Framework version 3 (CCFv3) is the standard reference atlas for the adult mouse brain. It parcellates the entire brain into ~1,327 structures at 10 um isotropic voxel resolution, using the ARA (Allen Reference Atlas) ontology for naming. |
There was a problem hiding this comment.
approximate to an exact number is something really only AI would do , so
| The Allen Mouse Brain Common Coordinate Framework version 3 (CCFv3) is the standard reference atlas for the adult mouse brain. It parcellates the entire brain into ~1,327 structures at 10 um isotropic voxel resolution, using the ARA (Allen Reference Atlas) ontology for naming. | |
| The Allen Mouse Brain Common Coordinate Framework version 3 (CCFv3) is the standard reference atlas for the adult mouse brain. It parcellates the entire brain into ~1,300 structures at 10 um isotropic voxel resolution, using the ARA (Allen Reference Atlas) ontology for naming. |
or
| The Allen Mouse Brain Common Coordinate Framework version 3 (CCFv3) is the standard reference atlas for the adult mouse brain. It parcellates the entire brain into ~1,327 structures at 10 um isotropic voxel resolution, using the ARA (Allen Reference Atlas) ontology for naming. | |
| The Allen Mouse Brain Common Coordinate Framework version 3 (CCFv3) is the standard reference atlas for the adult mouse brain. It parcellates the entire brain into (currently) 1,327 structures at 10 um isotropic voxel resolution, using the ARA (Allen Reference Atlas) ontology for naming. |
but is that the number?
|
|
||
| - Covers all species, making it the only option for organisms without a species-specific brain atlas | ||
| - Resolvable URIs via the OBO PURL system (e.g., `purl.obolibrary.org/obo/UBERON_0001954` for Ammon's horn) | ||
| - Formal bridge ontologies exist to the Allen Mouse Brain Atlas (`uberon-bridge-to-mba.owl`) and the Allen Human Brain Atlas (`uberon-bridge-to-aba.owl`) |
There was a problem hiding this comment.
where those .owl are? add hyperlinks?
| - Covers all species, making it the only option for organisms without a species-specific brain atlas | ||
| - Resolvable URIs via the OBO PURL system (e.g., `purl.obolibrary.org/obo/UBERON_0001954` for Ammon's horn) | ||
| - Formal bridge ontologies exist to the Allen Mouse Brain Atlas (`uberon-bridge-to-mba.owl`) and the Allen Human Brain Atlas (`uberon-bridge-to-aba.owl`) | ||
| - Well-established in the biomedical ontology ecosystem; used by ENCODE, Bgee, Phenoscape, and others |
There was a problem hiding this comment.
hyperlinks to those projects/acronyms
| **All other species:** Store UBERON IDs. As species-specific atlases become available (e.g., via BrainGlobe), add support for storing their IDs alongside UBERON. | ||
|
|
||
|
|
||
| ## Roll-out plan for dandi-cli |
There was a problem hiding this comment.
oh, by now I felt like reading user-oriented overview and thought that I am in -docs repo.
Then now we switched to "dandi-cli" details and I mentioned that we are in dandi-archive! Anything to develop in -archive actually here? should move to dandi-cli?
overall feels like a useful overview but I wonder if should be here and in such a verbose textual form -- hard to wrap head around what is actually steps done/wip/tobe done. Should we establish github project for this with individual issues etc assigned?
There was a problem hiding this comment.
I don't think this actually requires any changes to the dandi-archive repo specifically. I was just using this as a central place for design docs. I see there is a corresponding place in the CLI for design docs as well (https://github.com/dandi/dandi-cli/tree/master/docs/design), that has a doc from 2 years ago. We could move this over there if you really want, but I feel this is a cross-cutting discussion. It will require changes to dandi-cli, but there are also proposed changes to NeuroConv and the Atlas Explorer.
The whole point of this doc is to establish a strategy that will be rolled out over months across multiple PRs on several different repos. I know it's verbose, but it requires a lot of context to justify the nuanced per-species approach I'm recommending. The phases are meant to represent issues/PRs that can be opened once we agree on the strategy.
… anatomical ontologies documentation
|
|
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
For humans SNOMED CT (https://bioportal.bioontology.org/ontologies/SNOMEDCT) I know as the one frequently used and IIRC being adopted by DICOM standard (or some initiative within)... edit: it is quite "high level" as aims to describe full body, so would be not very applicable for brain data, BUT could be of better relevance for annotating data in BBQs/EMBER archive eg. on pose estimation etc; so might still be handy to keep in mind, mention, albeit not to recommend as the main one for humans. |
|
Additional note from while chatting with @dbkeator who pointed me to freesurfer, FSL an ANTs to UBERON mappings used for humans (we think). |
3 participants
Document the strategy for anatomical ontologies in DANDI, outlining goals, available systems, and implementation phases for dandi-cli and DANDI Atlas Explorer.