Review classification under "anatomical cluster"

[gouttegd]

It has already been noted elsewhere (https://github.com/obophenotype/cell-ontology/issues/1782) that several terms currently classified under anatomical cluster do not belong there. Time to review and fix those.

One of those terms at least is a cause of unsats as a result of that bogus classification: hard palate (UBERON:0003216).

[gouttegd]

Before proposing any changes to the classes below anatomical cluster, I’d like to make sure that there is a consensus about what an “anatomical cluster” actually is.

The term is currently defined as “an anatomical group that has its parts adjacent to each other”. It is a subclass of disconnected anatomical group, which is itself defined as “a material anatomical entity consisting of multiple anatomical structures that are not connected to each other”.

One question that the current definition raises is the meaning of “adjacent”. Should it mean that the parts making up the clusters are in physical contact with each other, or merely that they are close to each other?

If we take it to mean that the parts are in physical contact, then that would mean that they are connected, in which case the classification under disconnected anatomical group is wrong by definition.

For reference, FBbt’s definition for anatomical cluster is “an anatomical group whose component anatomical structures lie in close proximity to each other”. I may have a pro-FBbt bias here, but I believe FBbt’s definition is more correct – and incidentally, it is consistent with the definition of “cluster” in most dictionaries (Cambridge, Merriam-Webster, Collins).

I’d like to propose that Uberon follows FBbt on that point. Then we could decide if a given entity made of several parts is an anatomical cluster or something else as follows:

• Are the parts all connected? Then it’s some kind of anatomical structure.
• Otherwise it’s some kind of disconnected anatomical group.
  • Are the parts in close proximity? Then it’s a anatomical cluster. (I am aware that “close proximity” leaves room for interpretation; I am open to suggestions for a stricter criterion to apply here)
  • On the contrary, are the parts spread throughout the organism, yet still function together to fulfil a common function? Then it may be a non-connected functional system (e.g. endocrine system).

[cmungall]

I think these kinds of very abstract grouping classes should be modeled as properties or characteristics of some kind. See:

https://github.com/BFO-ontology/BFO/issues/221

but happy to align with fbbt as an intermediate step

[gouttegd]

In Uberon disconnected anatomical group is a subclass of 'has quality' some disconnected (PATO:0010001), and anatomical structure is a subclass of 'has quality' some 'maximally connected' (PATO:0010000).

So you would suggest to find (or create if they don’t already exist in PATO or elsewhere) more qualities to more finely distinguish the different types of entities? E.g., something like 'has quality' some 'closely grouped' or similar for anatomical cluster, if we follow FBbt’s meaning of the term?

[cmungall]

The PATO term seems a bit vague here. I think if we feel that clustered in the FBbt concept is a useful concept for AOs then it's worth making a PATO subclass?

It looks like many of the existing 'clusters' are essentially mereological sums / "set" terms. I think it's worth having a DP for these but I don't think that is the intent of the CARO class

[gouttegd]

Just to be clear, closely grouped does not exist in PATO, I just made it up for the example because it doesn’t look like there already is a quality in PATO that we could use to qualify “clusters”.

I’d be happy with a clustered quality instead, with an agreed upon definition.

[gouttegd]

Here’s the rest of the problems with the anatomical cluster hierarchy.

Barring any objection, I’ll submit a PR to fix the issues highlighted in the first two sections below (cross-references and subclasses that belong elsewhere). The other issues will require more time and more input.

Obsolete cross-references

Uberon’s anatomical cluster has a cross-reference to CARO’s anatomical cluster, which is obsolete and replaced with multi organ part structure, which has a different meaning than the one intended in Uberon for “anatomical cluster” (and which is already linked to Uberon’s own multi organ part structure.

Unless we happen to have a policy of keeping cross-references to obsolete foreign terms (e.g. for backwards compatibility?), the CARO cross-reference should be removed. And along with it, I would also suggest to remove all cross-references to what I would call “CARO term copies” – terms in other ontologies that mimic the obsolete CARO term, sometimes to the point that they use the same CURIE number (e.g. BILA:0000041, HAO:0000041, FZA:0001478...).

Cross-reference to FMA:49443 should be removed as well, since the FMA term has a very different definition (which is closer to multi organ part structure than to anatomical cluster).

Subclasses that belong elsewhere

A handful of classes can confidently be re-classified somewhere else than under anatomical cluster:

• accessory articulation -> move under multi organ part structure;
• glottis -> move under anatomical structure;
• hard palate -> move under multi organ part structure;
• soft palate -> move under multi organ part structure;
• head capsule -> move under [insect region of integument]() (mimic the FBbt classification; the term should also be rename to make it clearer it is intended for insects);
• insect analia -> move under anatomical structure (mimic the FBbt classification);
• juxtaglomerular apparatus -> move under anatomical structure, as per the definition;
• muscle spindle -> move under anatomical structure;
• lower urinary tract -> move under organ system subdivision;
• upper urinary tract -> move under organ system subdivision;
• ocular adnexa -> move under multi organ part structure;
• osseus cochlea -> move under anatomical structure;
• palatoglossal arch -> move under multi organ part structure;
• pharyngeal arch system -> move under multi organ part structure.

Under-defined, possibly taxon-specific classes

A handful of classes seem to come straight from taxon-specific ontologies, have little or no definition (making it difficult to assess whether they belong under anatomical cluster of not), and may have (if they are really taxon-specific) a name that is too generic:

• intervening zone;
• medial column;
• midline column;
• opercular series — the name “series” does suggest it is a cluster of some sort (like other similarly named classes such as infraorbital series or epibranchial series), but without any definition it's hard to be sure;
• pars inferior ear;
• pars superior ear;
• postzonal element;
• prezonal element;
• zonal area;
• ventricular musculature;
• visceromotor column;

Other various problematic classes

• entire embryonic mesenchym — definitely not a “cluster” under any reasonable definition, but unsure where it should go;
• exoskeleton — likewise;
• hypophysial region;
• proliferative region, coming from TAO, where it has as “exact synynoms” terms such as cell cycle, proliferation, cell division (to which my reaction is “err, what the hell…?”); ZFA defines it as “cluster of rapidly dividing cells”, which sounds like something that could be made more generic and used elsewhere;
• region of neural crest — the cross-referenced terms in ZFA, EMAPA, and TAO all represent the actual neural crest and not a region of it, and the definition also sounds like the term is referring to the entire neural crest, so it’s unclear what the term is really about;
• renal pelvis / ureter — a term intended to group the renal pelvis and the ureter — is such a grouping term really necessary? if so, it probably belongs under organ system subdivision;
• shell — again, not a “cluster”, but unsure where it should belong;
• subungual region — not a “cluster” and possibly redundant with nail bed;
• vasculature — not a “cluster” and possibly redundant with vascular system;
• dorsal fin pterygiophore — the definition makes it a “cluster” all right, but it seems that the name actually refer to the individual elements that make up the cluster (which seems confirmed by all the dorsal fin pterygiophore X subclasses);
• anal fin pterygiophore — likewise.

[github-actions[bot]]

This issue has not seen any activity in the past 6 months; it will be closed automatically one year from now if no action is taken.

[github-actions[bot]]

This issue has been closed automatically because it has not been updated in 18 months. Please re-open if you still need this to be addressed.