Should endothelial cell be a type of epithelial cell? (ditto for tissues in Uberon)

[dosumis]

@dosumis = I think no - most biologists would not group in this way - perhaps apart from some evo-devo folks studying the ancient evolutionary transitions (diploblast to triploblast?). Practically - this causes problems for classification. Epithelial cells types end up encompassing endothelial cell types due to automated classification, even though no-one making these terms for vertebrates would want that (AFAIK)

@emquardokus notes - endothelium and epithelium (in the strict sense) arise from different germ layers. Shouldn't that be enough to keep them separate?

Chris disagrees: Afaict no one agrees on the epithelium is endothelium issue. I'd be wary of making changes without thinking through implications. Esp multi species. Every time is see “true” as an adjective in ontologies i fear jagged inconsistent lattices

[ghost]

Additional comments from the CL/Uberon meeting on 28/3/2022:

Both Uberon and CL have this, so change needs to be coordinated.

We also have classification by germ layer, endo, ecto and meso-epithelium, with the latter being the parent class for endothelium (mirrored in CL).
In CL at least, these are maintained by hand. It seems likely that very many ecto-epithelial are directly under 'epithelial cell'.

Suggestions:

• make new broader terms for grouping (epithelial cell senso lato) subClassOf 'animal cell'? Or re-use an existing class for this?
• 'epithelial cell'
• Please add rdfs:comments to top-level affected terms (and seeAlso pointing to GitHub discussion) for tracking purposes

Thoughts:

• We stick with the status quo but make SOPs & edits to ensure we use ecto-epithelial terms where they should be used for specific classifications and logical definitions.

• We refactor to have ecto-epithelial cell have label 'epithelial cell' with ecto synonym and classifying comments.

• Who should we consult on solutions:
  People who rely on current classification
  Biological experts
  Biologists using the site

• Who can we realistically consult:
  HubMap

TO DO:

[ ] flesh out 2 solutions with aim to discuss again on next call.

[ghost]

Thoughts: would it be appropriate to simply remove superclass 'meso-epithelial cell' from 'endothelial cell'?
Is there a reference that supports this relationship?

Of note, from Wikipedia:
"The foundational model of anatomy, an index of terms used to describe anatomical structures, makes a distinction between endothelial cells and epithelial cells on the basis of which tissues they develop from, and states that the presence of vimentin rather than keratin filaments separates these from epithelial cells.[4] Many considered the endothelium a specialized epithelial tissue."[5]
PMID:22492947

[dosumis]

Suggestion:

• Most epithelial cells are almost certainly ecto-epithelial, but are not classified as such. We could fix this problem by correcting this - take everything classified as meso- or endo-eptihelial cell and subtract this from all epithelial cells. Change remaining to ecto-epithelial. Ditto epithelia terms in Uberon.

[github-actions]

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

[ghost]

Removing assignee. Once resources allow, the issue can be reassigned.

[dosumis]

This issue is fixed in CL by #2417 and associated PR. It has become urgent to fix for multiple HCA atlas projects where it is causing lots of unwanted inference. While in the broader scheme of things, it may be reasonable to group endothelium and epithelium, keeping them separate is both pragmatic and defensible.

I suggest adding some text to the comment field on 'endothelial cell' to clarify.

A similar disentangling is needed in Uberon - but the effects of the classification there do not seem to bleed through to CL at present.