How to logically define terms using 'anterior'?

[obophenotype-user]

Originally reported on Google Code with ID 59

1. can we post compose with BSPO anterior region term? 
For example:
Anterior scalloping of vertebral bodies
Proposed logical def:
has_part some (scalloped and ('inheres in' some ('anterior region' part_of some 'bony
vertebral centrum')))

2. For the term 'Anterior vertebral fusion'- we want to talk about group of vertebra
in anterior region. We have terms for cervical and thoracic- would we use 'inheres
in' union of cervical and thoracic? (note, there is not a lot of info about this term,
it comes from clinical synopsis on OMIM record, and there is not a textual definition)

Reported by vasilevs@ohsu.edu on 2014-10-06 18:21:34

[obophenotype-user]

Reported by vasilevs@ohsu.edu on 2015-02-12 05:36:48

• Status changed: New

[drseb]

@cmungall can you comment?

[cmungall]

OK, a lot to unpack here.

First some asides: Just because the HPO terms says "scalloped" it doesn't mean PATO:scalloped is the right term to use here.

The PATO term means that it looks like this:

We want 'concave here'

Additionally the 3 relevant HPO classes should have complete synonyms attached:

 HP:0004580 ! Anterior scalloping of vertebral bodies [SYNONYM: "Anterior scalloping vertebral bodies" (related)]
 HP:0004586 ! Biconcave vertebral bodies [SYNONYM: "Biconcave 'codfish' vertebrae" (exact)] [SYNONYM: "Biconcave vertebrae" (exact)] [SYNONYM: "Codfish vertebrae" (exact)] [SYNONYM: "Fish vertebrae" (exact)] [SYNONYM: "Scalloping of vertebral bodies" (exact)]
 HP:0005121 ! Posterior scalloping of vertebral bodies [SYNONYM: "Posterior vertebral body scalloping" (exact)]

Even PATO:concave is correct but not sufficient, for full N+S conditions we would need a class biconcave. But at some point the ROI for axiomatization drops off.

In general, axiomatizing highly specific spatial morphologies is one of those cases where it's hard, easy to get wrong, and the returns are low.

But if you did want to do it here, the pattern above is fine, but I would request a precomposed uberon class for simplicity

[cmungall]

Before even tackling this one we need a natural language def:

[Term]
id: HP:0004557
name: Anterior vertebral fusion
namespace: human_phenotype
xref: UMLS:C1969393
is_a: HP:0002948  ! Vertebral fusion

You assume anterior means along the A/P axis, which I would too, thinking of HOX genes etc. But is this really what it means? We should check. A/P terminology can be misleading.