UBERON:0000482 basal lamina of epithelium is defined as:
Acellular anatomical structure that consists of a thin sheet of fibrous proteins that underlie and support the cells of an epithelium.
Two CL classes assert that a cell is part_of it:
EquivalentClasses(CL:0002169, CL:0002167 and (part_of some UBERON:0000482)) # basal cell of olfactory epithelium
SubClassOf(CL:4033093, part_of some UBERON:0000482) # limbal epithelial stem cell of cornea
A cell cannot be part of an acellular ECM sheet. Both definition texts have it right and the axioms don't — CL:0002169 says the cell is "located on the basal lamina", CL:4033093 says it "resides at the basal layer". Adjacency was translated into parthood.
Contrast the correct pattern used elsewhere for the same idea:
EquivalentClasses(CL:0002187, CL:0000646 and (part_of some UBERON:0002025)) # 'stratum basale of epidermis'
stratum basale is a cell layer, so parthood holds. CL currently carries both encodings; these two terms are the only users of the wrong one.
Proposed fix
Swap BFO:0000050 part_of → RO:0002220 adjacent to in both axioms. RO:0002220 is already used 14× in cl-edit.owl, including as an EquivalentClasses differentia (e.g. CL:0005020, CL:0019021, CL:0002670), so this needs no new relation.
No class currently asserts adjacent to some UBERON:0000482, so no new inferences follow from the change.
Residual issue, deliberately not addressed here: "adjacent to the basal lamina" is also true of sustentacular cells in the pseudostratified olfactory epithelium, so CL:0002169's equivalence axiom remains over-inclusive. That is a question about what the class means rather than which relation it uses — tracked separately in #3700.
PR: #3704.
Note: these two are the only asserted violations. Computing the transitive part_of closure inherited over is_a finds seven in total — see #3701 for the full set and the two upstream UBERON axioms that cause the rest.
@cmungall drove the agent, checked its reasoning, and is accountable for its output.
Generated by Claude Code
UBERON:0000482 basal lamina of epitheliumis defined as:Two CL classes assert that a cell is
part_ofit:A cell cannot be part of an acellular ECM sheet. Both definition texts have it right and the axioms don't —
CL:0002169says the cell is "located on the basal lamina",CL:4033093says it "resides at the basal layer". Adjacency was translated into parthood.Contrast the correct pattern used elsewhere for the same idea:
stratum basaleis a cell layer, so parthood holds. CL currently carries both encodings; these two terms are the only users of the wrong one.Proposed fix
Swap
BFO:0000050 part_of→RO:0002220 adjacent toin both axioms.RO:0002220is already used 14× incl-edit.owl, including as an EquivalentClasses differentia (e.g.CL:0005020,CL:0019021,CL:0002670), so this needs no new relation.No class currently asserts
adjacent to some UBERON:0000482, so no new inferences follow from the change.Residual issue, deliberately not addressed here: "adjacent to the basal lamina" is also true of sustentacular cells in the pseudostratified olfactory epithelium, so
CL:0002169's equivalence axiom remains over-inclusive. That is a question about what the class means rather than which relation it uses — tracked separately in #3700.PR: #3704.
Note: these two are the only asserted violations. Computing the transitive
part_ofclosure inherited overis_afinds seven in total — see #3701 for the full set and the two upstream UBERON axioms that cause the rest.@cmungall drove the agent, checked its reasoning, and is accountable for its output.
Generated by Claude Code