Abnormality of lingual tonsil (transitively) a subclass of Abnormality of face

[cmungall]

hacking session with @julesjacobsen we noticed an oddity:

Is there a TPV on the left side here?

here it is as a tree:

* [i] HP:0000118 ! Phenotypic abnormality
    * [i] HP:0000152 ! Abnormality of head or neck
        * [i] HP:0000234 ! Abnormality of the head
            * [i] HP:0000271 ! Abnormality of the face
                * [i] HP:0000153 ! Abnormality of the mouth
                    * [i] HP:0031816 ! Abnormal oral morphology
                        * [i] HP:0000163 ! Abnormal oral cavity morphology
                            * [i] HP:0000157 ! Abnormality of the tongue
                                * [i] HP:0030809 ! Abnormal tongue morphology
                                    * [i] HP:3000076 ! Abnormality of lingual tonsil

Each individual link seems justifiable but transitively the link from tonsil to face doesn't seem right. I think the problem here is terms like oral cavity and mouth can be contextual

(btw this is the current asserted structure, not inferred)

[pnrobinson]

Probably mouth is not part of face and the face stops at the vermilion border! Need to figure out a better placement.

[pnrobinson]

UBERON says the oral cavity is part of mouth https://www.ebi.ac.uk/ols/ontologies/uberon/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FUBERON_0000167 It makes mouth a part of head (But not part of face). It is also a subdivision of the digestive tract.

My intuitive definition for mouth includes both the lips and the oral cavity, but this might just be the result of never having asked what anatomist think.

Any suggestion? @cmungall

[pnrobinson]

@cmungall @matentzn I think we need to revise uberon to fix this -- any progress?

[matentzn]

@rays22 Could you take this on please?

[pnrobinson]

@rays22 any update? Can we close this?

[pnrobinson]

@rays22 can we move forward?

[pnrobinson]

@rays22 @matentzn I am trying for inbox/issue zero before Summer vacation for sanity. We have several issues that have been open without activity for over a year, and I am guessing that the issue tracker is not working for them. Should we create a focussed tracker for logical definitions and move the issues over? I think this tracker wants to be more for new term request etc.

[rays22]

I have made a ticket in Uberon to reconsider the partonomy of the tongue and the oral cavity, which could potentially fix the inference issues with HP:3000076 Abnormality of lingual tonsil.

Alternatively, modifying the EQ of HP:0000163 Abnormal oral cavity morphology could fix the issue in HPO and uPheno without restructuring Uberon. Replace the characteristic of part of property chain with characteristic of for more intuitive phenotype inferences: * Suggested EQ:

[rays22]

Probably mouth is not part of face and the face stops at the vermilion border! Need to figure out a better placement.

1. I can confirm that UBERON:0000165 mouth is not part of UBERON:0001456 face. Revising the HPO subclass assertion that an Abnormality of the mouth is an Abnormality of the face would be part of the resolution of this issue.
2. I can confirm that UBERON:0001723 tongue is not part of UBERON:0000167 oral cavity. Revising the HPO subclass assertion that an 'Abnormality of the tongueis anAbnormal oral cavity morphology` would be the second part of the resolution of this issue.

In summary, this issue stems from subclass assertions in HPO, and it is not an issue of the Uberon class hierarchy. A computed phenotype class hierarchy that is based on Uberon would not show the inferences above.

[balhoff]

1. I can confirm that UBERON:0000165 mouth is not part of UBERON:0001456 face. Revising the HPO subclass assertion that an Abnormality of the mouth is an Abnormality of the face would be part of the resolution of this issue.

However Uberon does say that every face has_part some mouth. So any instance model in which there is a face (like say a human) would imply that the mouth is part of the face (inverse properties). So the HPO subclass relationship is consistent with this.

[rays22]

1. I can confirm that UBERON:0000165 mouth is not part of UBERON:0001456 face. Revising the HPO subclass assertion that an Abnormality of the mouth is an Abnormality of the face would be part of the resolution of this issue.

However Uberon does say that every face has_part some mouth. So any instance model in which there is a face (like say a human) would imply that the mouth is part of the face (inverse properties). So the HPO subclass relationship is consistent with this.

Good point @balhoff . There is a new ticket in Uberon to make the lip ( (located at the opening of the mouth) part of the face in Uberon. This should allow us to drop the face 'has part' some mouth' assertion in Uberon, because the lips would represent the mouth-specific parts that are being visible parts of the face in contrast to mouth structures that are deeper in the mouth.

[rays22]

This is to summarise some of the 'part of' and 'has part' relationships in Uberon resulting from Uberon issue #3311.

• The lingual tonsil is no longer part of the face (transitively) in Uberon, and the
• face does not have the mouth as its part.
• Now the lip is asserted to be part of the face in Uberon.
• The lip remains also part of the mouth, and the
• lingual tonsil also remains part of the mouth in Uberon.

In summary, automatic reasoning should no longer infer that the lingual tonsil is part of the face. However, there is still at least one subclass assertion in HPO that needs to be revised to avoid inferring that an 'Abnormality of the lingual tonsil' is also an 'Abnormality of the face' phenotype.

[matentzn]

Thank you @rays22 for working on this and seeing this through! Much appreciated!