Replace the use of innervates for ganglia with `extends_fibers_into`

[anitacaron]

• I think we could replace the use of innervates for ganglia with 'extends_fibers_into', preserving use of innervates for neuron projection bundles (tracts and nerves) - although I guess that might lose the sense that those fibers synapse to targets in the object of the relationship.

Originally posted by @dosumis in https://github.com/obophenotype/uberon/issues/2495#issuecomment-1187364719

[shawntanzk]

@anitacaron - came up again in https://github.com/obophenotype/uberon/pull/2615 -> I have made this change there

[cmungall]

I agree with this change and given the problems this was causing let's try and get these fixes in sooner!

[ghost]

Thanks for sharing the following specific example, @anitacaron.

otic ganglion EquivalentTo cranial ganglion and (innervates some parotid gland) and (extends_fibers_into some glossopharyngeal nerve)

Your proposal is to change this to: otic ganglion EquivalentTo cranial ganglion and (extends_fibers_into some glossopharyngeal nerve)

The result of this change infers that the 'otic ganglion' is equivalent to the 'glossopharygeal ganglion'. This is a region of complicated anatomy, but I am not sure that this is true. If it is indeed true (don't think it is, but would need to review to confirm), then why are these separate classes instead of exact synonyms?

CC @dosumis

[shawntanzk]

defs not ok - we don't allow infered equivalent -> will look into this. Thanks for bringing it up :)

[anitacaron]

After thinking, again and again, the solution would be this:

otic ganglion EquivalentTo cranial ganglion and (extends_fibers_into some parotid gland) and (extends_fibers_into some glossopharyngeal nerve)

Then it results into:

Do you think this is good, @bvarner-ebi?

[ghost]

Thanks for sharing, @anitacaron.

I am curious about the object property 'extend_fibers_into'. Based on the text of the label alone (I do not see a definition), and searching a few other relationships in Uberon, this suggests that fibres are originating from object X (in this case the 'cranial ganglion') and extending into object Y (the 'glossopharyngeal nerve'). In this example, it is opposite to the anatomy. The 'otic ganglion' receives fibres from the 'glossopharyngeal nerve'.

My question is: does 'extend_fibers_into' denote some sort of afferent/efferent direction? If not, then this axiom seems OK, although I recommend adding a text definition to 'extends_fibers_into" to make that clear. On the other hand, if 'extend_fibers_into' denotes direction and/or the presence of a synapse, this would need to be remodelled and the proposed solution may need to be reconsidered.

And, so we are on the same page, I had a look at the linked comment:

(I'm sure very many ganglia also have whole cells)

A ganglion contains nerve cell bodies, which are portions of cells. My understanding is that ganglia do not contain whole cells, so if the above statement is true, an example would be helpful.

[ghost]

EDIT to comment above. I originally had:

FMA also shows 'otic ganglion' to be a child of 'glossopharyngeal ganglion', so there is alignment with FMA.

I arrived to this browser in the link after searching for an FMA browser, but this does not appear to be an FMA browser.

[ghost]

@anitacaron, would the following work for you since I have not found evidence that the 'otic ganglion' is a subclass of 'glossopharyngeal ganglion'?

[shawntanzk]

does 'extend_fibers_into' denote some sort of afferent/efferent direction

Not that I am aware of, but as you mentioned, lack of definition - should really try to get that into RO, though not sure how they would feel about that vs innervates.

In this example, it is opposite to the anatomy. The 'otic ganglion' receives fibres from the 'glossopharyngeal nerve'.

I guess receives fibres does however suggest that glossopharyngeal nerves extends fibers to otic ganglion, though that does not mean the opposite isn't true - will look into this.

A ganglion contains nerve cell bodies, which are portions of cells. My understanding is that ganglia do not contain whole cells, so if the above statement is true, an example would be helpful.

I'm not 100% sure, but I'm guessing that's more hedging than anything, I think it should be alright to just go to with UBERON:0005162 multi-cell-part structure but I see that it has already been changed to anatomical structure which isn't wrong. I guess its to be careful in case there are exceptions, which seems often in the nervous system. Also I'm not 100% sure if we would consider glial cells to be part of the ganglion, which then, chances are there are full glial cells in the structure.

[shawntanzk]

have not found evidence that the 'otic ganglion' is a subclass of 'glossopharyngeal ganglion'?

100% agree with this. Perhaps figuring out the relation with Tympanic nerve would model it more accurately model this. See images below: EDIT: relation with lesser petrosal nerve

from https://teachmeanatomy.info/head/cranial-nerves/glossopharyngeal-nerve/ from https://memorang.com/flashcards/85882/003B+-+Temporal+Region from here

[shawntanzk]

@anitacaron, would the following work for you since I have not found evidence that the 'otic ganglion' is a subclass of 'glossopharyngeal ganglion'?

btws personally agree with this solution for now, though would like to breakdown the subclass of cranial ganglion more

Would however be good to get a signoff by @dosumis before proceeding

[ghost]

@shawntanzk, thanks for aggregating these images.

A caveat, though... there are also neurones that pass through the otic ganglion that do not synapse in the ganglion, which are not shown in these illustrations (ref):

"The origin of sympathetic neurons associated with the otic ganglion arises from the plexus around the middle meningeal artery. These plexus convey postganglionic fibers arising from the superior cervical ganglion. The postganglionic sympathetic fibers pass through the otic ganglion without relay and run together with the parasympathetic fibers in the auriculotemporal nerve to provide sympathetic innervation to the parotid gland blood vessels.[2]"

And further evidence to divorce 'glossopharyngeal nerve' from 'otic ganglion' (PMID:3446656):

"In the rat there is no direct connection between the glossopharyngeal nerve and the otic ganglion, as is the case in man. However, a neural connection between the facial nerve and the otic ganglion and a further connection linking the glossopharyngeal with the facial nerve through which the salivatory axons may pass to reach the otic ganglion is described."

[dosumis]

Just skimming this, but

(a) I'd avoid an equivalent class definition for optic ganglion unless there's a really strong need to autoclassify under it and you're really sure it's safe. subClassOf restrictions are much safer.

(b) I think the definition of 'extends fibers into' should avoid cases like this: "there are also neurones that pass through the otic ganglion that do not synapse in the ganglion"

suggested def:

X extends_fibers into Y iff there exists some neuron(n) n has_soma_location(RO:0002100) x and n 'has synaptic IO in region'(RO:0013001) Y

I think it would be safe to add this property chain as I believe has_soma_location has domain neuron.

has_soma_location o 'has synaptic IO in region' -> 'extends_fibers_into'