PATO:Absent
From NCBO Wiki
This is a currently open tracker item:
Contents |
[edit] The problem
Currently (18Nov2006), PATO has the following structure for 'absent':
is_a PATO:0001241 quality of continuant
is_a PATO:0001237 monadic quality of continuant
is_a PATO:0000068 qualitative
is_a PATO:0000070 presence
is_a PATO:0000462 absent
Under current practice, when annotating a 'wingless' phenotype, a curator would state (eg via Phenote):
E= fly_ao:wing Q= PATO:absent
However, this is problematic. "absence" is not a true quality since it is impossible for an instance of any entity to not exist. PATO, following OBO Foundry principles, defines qualities that can inhere_in an actual existing entity. The above annotation literally states "There is a wing which is not there".
There is a second more practical problem in that the annotation method above does not allow us to easily state "spermatocyte devoid of asters". We can say
E= GO:aster Q= PATO:absent
but we lose the fact the aster is missing specifically from the spermatocyte.
We can get round this by Post-coordinating the entity term aster of spermatocyte:
E= GO:aster^part_of(CL:spermatocyte) Q= PATO:absent
(note that post-coordination can now be done easily through the UI of phenote)
But the ontological problem still remains. This is not a moot theoretical point - this affects the kinds of inferences we can make over annotations
[edit] The solution
Any solution must balance ontological correctness with simplicity, ease-of-use and satisfying a biologists intuitive sense of correctness. I first present the preferred way of doing things, and then introduce a compromise to make current curations backwards compatible.
[edit] New PATO term: lacking_part
This is a relational quality (it can be argued that the appropriate place for this is in the OBO Relation ontology - however, it's useful to also keep this term in PATO). A relational quality takes on an additional argument/parameter; for example:
E= fly_ao:thorax Q= PATO:lacking_part E2= fly_ao:wing
'lacking_part' is a relational quality that inheres in an existing entity (eg the thorax) and is directed towards an additional entity type (in this case 'wing').
Of course, now we have 3 slots to fill in rather than 2 this option may not seem to appealing to a busy curator. See the compromise section, next.
It is also perfectly valid to create this annotation:
E= fly_ao:organism Q= PATO:lacking_part E2= fly_ao:wing
The two can be annotations can be determined to be equivalent (provided wing is always part of thorax in the canonical fly, which it is).
In fact, one could simply state
Q= PATO:lacking_part E2= fly_ao:wing
And have the primary entity implied.
This is crucially different from the nonsensical annotation introduced at the start of this document; the relationship between Q and E is one of inherence - the E bears the Q, and this is true of the actual instances described in the annotation. In contrast, the Q is directed towards the E2, which can be a type.
From a purely practical perspective, this annotation methodoogy is preferable, as it allows us to state:
E= CL:spermatocyte Q= PATO:lacking_part E2= GO:aster
See also PATO:Relational_qualities
[edit] A compromise to simplify annotation
Many existing annotations have been made to PATO:absent; it seems that some curators would prefer to continue annotating this way. In particular, evolutionists creating character matrixes will have a hard time thinking of presence/absence being different from real qualities such as square/round.
My proposal is to retain PATO:0000462 "absent", with the same ID, but to move it to a separate section of the ontology, perhaps under a sister term to "quality", such as "pseudoquality" or "information attribute". We would introduce a completely separate term for "lacking_part", a relational quality.
It is then relatively trivial for software to convert
E= <e> Q= PATO:0000462
to
E= CARO:organism Q= PATO:lacking_part E2= <e>
The conversion could happen at edit time, when loading into the local database (eg MOD), or perhaps only when loading into a database capable of sophisticated reasoning (eg OBD). The decision would be up to the organisation to which the curator belongs.
[edit] Alternate solution
It has also been suggested (obo-phenote list, Nov22 2006) that present/absent could be placed in a separate 'Tag' field. This is not recommended, Tag was set up as a temporary placeholder whilst a theory of what consititutes abnormal is worked out. See PATO:Main_Page#Pheno-syntax
[edit] Similar problems
[edit] Counts
A similar analysis to the one above can be followed for PATO:count
is_a PATO:0001241 quality of continuant
is_a PATO:0001237 monadic quality of continuant
is_a PATO:0001226 quantitative
is_a PATO:0000053 count
A similar solution would have relational qualities for having_extra_part(s), having_missing_part(s). These would take two extra arguments - an E2 and a count
[edit] Lack of function
Active discussion on mail list will be highlighted here later
[edit] Absence of a process
The favoured way of specifying this:
E= organism_development Q= lacks_part E2= wing_development
The acceptable substitute:
E= wing_development Q= absent
The two can be inter-converted
[edit] Related material
There are some slides that may be relevant in this presentation: Media:pato-hinxton-2006.ppt
Pages using the relation “Absent”
Showing 0 pages using this relation.
