missing human teeth when comparing FMA to UBERON

[wdduncan]

In order to use UBERON in the OHD, I need at least a complete set of human teeth (i.e., dentition).

I did a comparison between the FMA and UBERON, and found that UBERON was missing 52 (e.g., FMA:55696 "Right upper third secondary molar tooth"). The full results can be seen here.

The instructions for how I created the file are found on the README.

[cmungall]

See this issue

• hubmap requirements for bilateralogs and serialogs #2043

for a general discussion of how we handle requests for lateral subtypes.

Previously our recommendation has been to postcoordinate laterality for structures that exhibit lateral symmetry. We can always revisit this.

However we need to do this intentionally being aware of the maintenance implementation, and not do this in a piecemeal fashion

Here is what I mean:

A tooth has parts. E.g. cingulum. We already precoordinate classes like cingulum of [upper,lower]{molar, incisor}[NUMBER] etc. If we precoordinate left and right for all tooth subclasses, do we also precoordinate the lateral parts?

1. if we do not, then we have the ragged lattice problem
2. if we do, then we need a strategy for inferring the part-ofs so we don't manually assert the partonomy

For 2 we have to be careful about the DPs, see #2696

[wdduncan]

Previously our recommendation has been to postcoordinate laterality for structures that exhibit lateral symmetry.

Sorry. I was not aware of this recommendation.

I can post-coordinate in the OHD, although then I'm faced with the choice of using the FMA classes for laterality or to just create new OHD classes.

[cmungall]

not sure what you mean.. if you can postcoordinate then you just need the base classes

if you don't have the optional of postcoordinating then either we precoordinate in uberon or you use FMA

[wdduncan]

not sure what you mean.. if you can postcoordinate then you just need the base classes

Yes. I can create specific tooth classes in the OHD. E.g:

ohd:'upper right wisdom tooth' rdfs:subClassOf uberon:'upper third secondary molar tooth'

I can also import FMA:55696 (Right upper third secondary molar tooth) and define it as subclass of the UBERON class. E.g.:

fma:'Right upper third secondary molar tooth' rdfs:subClassOf uberon:'upper third secondary molar tooth'

If I take the FMA route, there may be a few drawbacks:

• The FMA is no longer maintained (as far as I can tell)
• I have to decide if I keep the FMA axioms (e.g., 'constitutional part' some 'Cavity of right upper third molar tooth')
• I might be changing the semantics of the FMA class (see screenshot)

[wdduncan]

A tooth has parts. E.g. cingulum. We already precoordinate classes like cingulum of [upper,lower]{molar, incisor}[NUMBER] etc. If we precoordinate left and right for all tooth subclasses, do we also precoordinate the lateral parts?

I understand how this can cause maintenance issues :(
For what it is worth, the FMA does include a number of lateral specific tooth parts. E.g.:

The OHD is clinically biased. So, the lateral distinctions between teeth are needed (e.g., need to represent which wisdom tooth was extracted). In the clinical records I've worked with, I haven't seen tooth parts defined by laterality. For example, if the distal surface of the left lower canine was restored, there is no special code for the lower left canine distal surface.

If I assert the FMA classes to be subclasses of the UBERON classes am I committing the cardinal sin of axiom injection?
Given that the FMA is not active, I am leaning towards simple creating the tooth classes in the OHD. This is more work for me in the short term, but may work out better in the long run.

[cmungall]

would you be up for adding the terms to uberon via dosdps?

[wdduncan]

Yes. I could do that. I'd have to brush up on dosdps :)

[shawntanzk]

Hi @wdduncan - I'm assigning this ticket to you, hope that is ok :)

[wdduncan]

@shawntanzk fine with me :)

[wdduncan]

@shawntanzk What is the best way to add this using the dosdp-patterns?
In the dosdp-patterns directory I see some tooth patterns: lower_tooth_pattern, upper_tooth_pattern.

Should I add to these or create a new pattern (e.g., human_tooth_pattern.yaml)?

[shawntanzk]

Ah I did remember @megbalk doing some work with this - she would be the best person to talk to regarding this (I don't know the patterns and biology as well as I assume she would). Happy to help but she is probably a better first point :)

[uberon]

We don't add species-specific terms, if I understand correctly what you need is laterialized teeth. We could have a pair of general laterilization patterns, eg left_X and right_X

…

On Thu, Oct 27, 2022 at 8:13 AM Bill Duncan ***@***.***> wrote: @shawntanzk <https://github.com/shawntanzk> What is the best way to add this using the dosdp-patterns? In the dosdp-patterns <https://github.com/obophenotype/uberon/tree/master/src/patterns/dosdp-patterns> directory I see some tooth patterns: lower_tooth_pattern <https://github.com/obophenotype/uberon/blob/master/src/patterns/dosdp-patterns/lower_tooth_pattern.yaml>, upper_tooth_pattern <https://github.com/obophenotype/uberon/blob/master/src/patterns/dosdp-patterns/upper_tooth_pattern.yaml> . Should I add to these or create a new pattern (e.g., human_tooth_pattern.yaml)? — Reply to this email directly, view it on GitHub <#2692 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAGZRCEPOANYDAWYRDDQ3NTWFKL3NANCNFSM6AAAAAARPHJH3M> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[meghalithic]

We can add right and left to the patterns and just ignore right or left if not used?

[cmungall]

the existing system doesn't work well with unspecified fields. Others can correct me if I'm wrong but I think the best way to do this is two new patterns, two new TSVs

Later on we can explore using linkml-owl which allows a single yaml object for a laterializable structure to be in one record with all 3 ids

[wdduncan]

@cmungall do mean two new patterns for the right/left teeth? E.g., right_tooth_pattern.yaml.

Also, I'm not following some of the naming conventions. E.g.:

• upper molar 1
  • upper first molar primary molar tooth
  • upper first secondary molar tooth

Do we want the parent to be named upper first molar tooth? Also, do we need the "tooth" modifier? Are there other molars that aren't teeth? All my google searches for "molar anatomy" returned teeth.

Also, I noticed that upper first secondary molar tooth is defined using the axiom:

'molar tooth 1'
 and ('part of' some 'upper jaw region')
 and ('part of' some 'secondary dentition')

dentition is defined as:

'anatomical collection'
 and ('has member' some 'calcareous tooth')

So, shouldn't we be using the member of relation for defining teeth (e.g., 'member of' some 'secondary dentition')?

[cmungall]

Also, do we need the "tooth" modifier? Are there other molars that aren't teeth

Nope :-) I leave it to you what are the best conventions for primary label. Of course, we can accommodate others in synonyms.

Note that we strive for consistency. The base class should be called "molar tooth" as "molar" is not going to be unique across OBO (chemistry).

[cmungall]

Also, I noticed that upper first secondary molar tooth is defined using the axiom:

'molar tooth 1'
and ('part of' some 'upper jaw region')
and ('part of' some 'secondary dentition')

This is a bad logical definition. See #2696

[cmungall]

dentition is defined as:

'anatomical collection'
and ('has member' some 'calcareous tooth')
So, shouldn't we be using the member of relation for defining teeth (e.g., 'member of' some 'secondary dentition')?

good point.

[wdduncan]

Note that we strive for consistency. The base class should be called "molar tooth" as "molar" is not going to be unique across OBO (chemistry).

Agreed :)

Wondering about the definition for molar tooth 1:

The molar tooth of the upper or lower jaw that is phylogenetically number 1.

Do you know of a reference about this? I do not know if phylogenetically number 1 aligns with what dentist call the "first molar". E.g., In the figure below, I do not know if the teeth labeled "First Molar" are called that b/c they are the first molar encountered when counting teeth from front to back, or if this order aligns with phylogenetic order of how these teeth came to be. (P.S. The innermost numbers are the ADA labels teeth; the outermost numbers are the FDI labels for teeth; the little diagrams between the numbers come from the Palmer system.)

[cmungall]

@skansa @megbalk any thoughts on this? should we continue to group by homology?

[wdduncan]

Looking at this figure in this article, I am tempted to say that a "first molar" is type of molar tooth 1. But, I will defer to those who know more about tooth evolution (or evolution in general).

[wdduncan]

Can someone give me permissions to create new branches and do a PR w/o having to fork UBERON?

[cmungall]

you should have an invite

[wdduncan]

Thanks @cmungall !

[wdduncan]

since I'm adding specifically human teeth, do you want me to use the present in taxon OP? E.g.:

upper secondary right third molar present in taxon Homo sapiens

Or is there another/better way to tag the tooth type as being human specific?

[cmungall]

uberon is a multi-species ontology, there are very few species specific terms in it. you should make these as general as possible

[wdduncan]

@cmungall the present in taxon annotation seems pretty general to me:

S present_in_taxon T if some instance of T has some S. This does not means that all instances of T have an S - it may only be certain life stages or sexes that have S

As far as I can tell, it doesn't commit to a tooth type being only in humans.

There are uses of present in taxon for humans. E..g.:

adipose tissue present in taxon Homo sapiens

[cmungall]

@wdduncan so sorry I didn't read your original comment closely enough, I read it as you wanting to put only-in-taxon

yes, please add present-in-taxons! thanks 🙏 🙏 🙏