incorrect taxon constraints

[lreiser]

Uberon term
cuticle development
regulation of epidermal cell differentiation
stem cell differentiation

and others....

Bug description

this is prompted by a (now very old) ticket in the GO where many plant annotations are being flagged as taxon constraint violations because of associations to UBERON terms. Pascale Guadet requested we open a ticket her in Uberon.

geneontology/go-ontology#30555

Reviewing our latest snapshot for GO rule violations, we noticed a bunch of new taxon violations for old Arabidopsis annotations to these terms. (sample only, there are more)

cuticle development
regulation of epidermal cell differentiation
stem cell differentiation

Was there a mass mapping of some older GO terms to UBERON terms for logical defs and the GO terms then acquired their taxon constraints? Can we revisit some of these and address the taxon constraint issues?

[cmungall]

@dragon-ai-agent please analyze the use of the term cuticle across metazoa, suggest more appropriate specific classes for different clades.

this is what we have now:

id: UBERON:0001002
name: cuticle
def: "Tough but flexible, non-mineral outer coverings of an organism, or parts of an organism, that provide protection. Cuticles are non-homologous, differing in their origin, structure and chemical composition[WP]." [http://sourceforge.net/tracker/?func=detail&aid=3348965&group_id=36855&atid=440764, Wikipedia:Cuticle]
comment: EDITOR_NOTE TODO - check cuticle vs exoskeleton. Note that GO has a very generic definition of cuticle. See also arthropod-anatomy ontology. Note in ncit Cuticle is a plant part
subset: grouping_class
subset: non_informative
subset: uberon_slim
synonym: "cuticula" RELATED [BTO:0001600]
synonym: "cuticular exoskeleton" RELATED []
xref: BTO:0001600
xref: Wikipedia:Cuticle
is_a: UBERON:0000476 ! acellular anatomical structure
disjoint_from: UBERON:0004882 ! eponychium
relationship: in_taxon NCBITaxon:6072 {source="Rieger - Evolution of the Cuticle in the Lower Eumetazoa"} ! Eumetazoa

[ai4c-agent]

🤖 Working on it...

Follow along: View workflow run

— @dragon-ai-agent

[ai4c-agent]

Analysis of UBERON "cuticle" across Metazoa

1. Current state in UBERON

ID	Name	Taxon	Notes
UBERON:0001002	cuticle	in_taxon Eumetazoa (NCBITaxon:6072)	grouping_class, non_informative, uberon_slim; defined as explicitly non-homologous
UBERON:0001001	chitin-based cuticle	in_taxon Arthropoda (NCBITaxon:6656)	equivalent: cuticle and composed_primarily_of chitin
UBERON:0003049	collagen and cuticulin-based cuticle	in_taxon Nematoda (NCBITaxon:6231)	taxon_notes: "currently restricted to nematodes, but this may be too strict"
UBERON:0003201	exocuticle	Arthropoda	Layer of arthropod procuticle
UBERON:0003202	endocuticle	Arthropoda	Layer of arthropod procuticle
UBERON:0002076	cuticle of hair	(implicit vertebrate)	Mammalian hair shaft outer cell layer — a total misnomer semantically, shares only the name
UBERON:0006693	obsolete protein-based cuticle	—	Obsolete, see also GO:0007592

2. Why the current model causes the GO problem

GO's cuticle development (and relatives) have logical definitions that reference UBERON:0001002 cuticle. Because UBERON:0001002 carries in_taxon Eumetazoa, any GO term that uses it inherits a constraint that forbids plant annotations — producing the Arabidopsis taxon violations reported in geneontology/go-ontology#30555.

Note also that the text definition of UBERON:0001002 explicitly states "Cuticles are non-homologous, differing in their origin, structure and chemical composition" — which already acknowledges that the class is a heterogeneous grouping, not a natural kind. The in_taxon Eumetazoa is really an editorial choice to delegate plant cuticle to PO, not a biological statement about homology.

3. Biological survey: cuticle-bearing metazoan clades

Ecdysozoa (all defined by periodic molting of a chitinous/proteinaceous cuticle)

• Arthropoda — chitin + sclerotized cuticular proteins (resilin, cuticulins). ✅ UBERON:0001001
• Tardigrada — chitinous cuticle with sclerified plates; molted. ❌ missing (currently would fall under arthropod cuticle, which is wrong)
• Onychophora — thin chitinous cuticle with papillae; molted. ❌ missing
• Nematoda — collagen + cuticulin + glycoproteins; molted four times in development. ✅ UBERON:0003049
• Nematomorpha — thick chitinous cuticle; molted. ❌ missing
• Priapulida / Kinorhyncha / Loricifera (Scalidophora) — chitinous cuticle bearing scalids or lorica; molted. ❌ missing

Lophotrochozoa (non-molted cuticles; continuously secreted)

• Annelida — collagenous cuticle, thin and non-molted. ❌ missing (important — annelid cuticle is collagen-based but structurally distinct from nematode cuticle)
• Rotifera — intracytoplasmic lamina: proteinaceous layer inside syncytial epidermal cells, not a typical extracellular cuticle. ❌ missing
• Mollusca — periostracum, organic conchiolin outer layer of shell; cuticle-like but usually not called "cuticle". Also distinctive cuticle on the foot/mantle in some taxa. ❌ missing
• Chaetognatha, Entoprocta, Bryozoa — thin cuticles in various locations. Usually not a priority.
• Platyhelminthes — no cuticle; surface is a syncytial tegument (cellular, not acellular).

Deuterostomia

• Hemichordata — delicate cuticle over proboscis and collar in some enteropneusts. Minor.
• Echinodermata, Chordata — no true body cuticle.
• Vertebrata — the only vertebrate structure bearing the word "cuticle" (UBERON:0002076 cuticle of hair) is a specialized keratinized cell layer on the mammalian hair shaft; it is not homologous or structurally comparable to invertebrate cuticles and should arguably be renamed or at minimum kept logically disjoint from UBERON:0001002.

Non-bilaterians (Cnidaria, Ctenophora, Porifera, Placozoa)

• No cuticle. The current in_taxon Eumetazoa constraint is therefore slightly too broad — a tighter constraint would be in_taxon Bilateria.

4. Recommendations

4a. Re-examine the taxon constraint on UBERON:0001002 (root cause of GO issue)

Three realistic options, from least to most disruptive:

1. Downgrade to present_in_taxon / remove strict in_taxon: UBERON:0001002 is already flagged grouping_class and non_informative. Because the class is by construction non-homologous, a hard in_taxon is dubious. Removing it (or replacing with never_in_taxon for specific excluded clades like Porifera) lets GO's cuticle development apply cross-kingdom without false violations.
2. Keep Eumetazoa constraint, but fix GO side: GO should either (a) create cross-kingdom cuticle development without the UBERON logical def, or (b) split into clade-specific GO terms that reference the clade-specific UBERON cuticle classes. This pushes the fix to GO.
3. Tighten to Bilateria (more correct biologically) and add a parallel plant cuticle grouping via PO cross-reference — doesn't solve the GO problem but is more accurate.

Recommendation: option 1 — remove in_taxon Eumetazoa from UBERON:0001002 and rely on the clade-specific subclasses for taxon constraints. Add a comment noting that the class is a non-homologous grouping and plant cuticle is represented in PO.

4b. Broaden / correct existing clade-specific classes

• UBERON:0001001 chitin-based cuticle: change in_taxon Arthropoda → in_taxon Panarthropoda (NCBITaxon:88770) to properly include Tardigrada + Onychophora, or create a parent ecdysozoan chitinous cuticle at the Ecdysozoa level (NCBITaxon:1206794) and put Arthropoda, Tardigrada, Onychophora, Scalidophora, Nematomorpha as sub-taxa of it. The latter is cleaner because it matches the molting character that defines Ecdysozoa.
• UBERON:0003049 collagen and cuticulin-based cuticle: keep in_taxon Nematoda as is; the taxon_notes caveat can be removed since annelid cuticle is structurally distinct enough to warrant its own class (see below). Consider renaming to nematode cuticle for clarity.
• UBERON:0002076 cuticle of hair: mark disjoint_from UBERON:0001002, or rename (e.g. hair shaft cuticular layer) to avoid confusion. It does not belong under the invertebrate cuticle hierarchy.

4c. Proposed new clade-specific classes

Proposed name	Taxon	Logical def / parent	Priority
annelid cuticle	Annelida (NCBITaxon:6340)	cuticle and composed_primarily_of collagen; not molted	High — fills a real gap in Lophotrochozoa
tardigrade cuticle	Tardigrada (NCBITaxon:42241)	sub of chitin-based cuticle	Medium
onychophoran cuticle	Onychophora (NCBITaxon:2074087)	sub of chitin-based cuticle	Medium
scalidophoran cuticle	Scalidophora (NCBITaxon:2008202)	sub of chitin-based cuticle	Low (small clades)
nematomorph cuticle	Nematomorpha (NCBITaxon:33310)	sub of chitin-based cuticle	Low
rotifer cuticle (intracytoplasmic lamina)	Rotifera (NCBITaxon:10190)	sub of cuticle with note on intracellular location	Low
ecdysozoan cuticle (optional grouping)	Ecdysozoa (NCBITaxon:1206794)	sub of cuticle; has_quality molted	Medium — captures the molting synapomorphy

4d. Don't forget chitin-based cuticle development on the GO side

Note there's already GO:0040003 chitin-based cuticle development (taxon-constrained to Arthropoda via UBERON:0001001). If Uberon creates ecdysozoan cuticle / annelid cuticle, GO can follow with matching cuticle development subclasses and the constraint issue evaporates for cuticle-related processes.

5. Suggested next step

My recommendation, in order of impact:

1. Immediate: Remove in_taxon Eumetazoa from UBERON:0001002 (or narrow to never_in_taxon Porifera/Placozoa/Ctenophora). This single change fixes the Arabidopsis violations without requiring any GO changes.
2. Short term: Add annelid cuticle (highest biological value; real lophotrochozoan class currently missing).
3. Short term: Broaden UBERON:0001001 to Panarthropoda (or create an ecdysozoan cuticle parent).
4. Optional: Add tardigrade/onychophoran/scalidophoran/nematomorph cuticle sub-classes as sub-classes of the broadened chitin-based / ecdysozoan cuticle.

Happy to open a PR implementing any subset of these. Please confirm which direction you'd like — in particular, whether to go with option 1 (simplest fix: drop the taxon constraint on UBERON:0001002) or option 2 (GO-side fix) for the root cause, and which new clade-specific classes you want created.

---

🤖 Generated by @dragon-ai-agent

• Model: claude-opus-4-7
• Agent harness: claude-code
• Triggered by: @cmungall
• Run: View workflow run

[cmungall]

@dragon-ai-agent please go ahead with your main recommendation and relax the TC. IMO it renders cuticle uselessly abstract but it causes least disruption. Add the grouping_class subset.

please make separate PRs improving the invertebrate cuticle representation.

[ai4c-agent]

🤖 Working on it...

Follow along: View workflow run

— @dragon-ai-agent

[ai4c-agent]

Done — opened #3695 implementing your main recommendation:

• Removed in_taxon NCBITaxon:6072 (Eumetazoa) from UBERON:0001002 (cuticle)
• subset: grouping_class was already present, so no change needed there
• Updated the comment to record the rationale (heterogeneous non-homologous grouping; constraint removed because it caused downstream taxon violations like the Arabidopsis ones in geneontology/go-ontology#30555) and to point users at clade-specific subclasses
• Added a term_tracker_item linking back to this issue

I'll file the invertebrate cuticle improvement PRs separately, starting with annelid cuticle (highest-value missing class) and the Panarthropoda broadening of chitin-based cuticle.

---

🤖 Generated by @dragon-ai-agent

• Model: claude-opus-4-7
• Agent harness: claude-code
• Triggered by: @cmungall
• Run: View workflow run