[Obo-relations] original_part_of is not transitive
Stefan Schulz
>
> Here's another example, going back to hearts:
>
> Consider our surgically removed heart. Let's say the extracted heart is kept
> alive, and we induce development of an entirely new part such as a
> ventricle, perhaps by stem cell therapy. The new ventricle is undoubtedly an
> original part of the heart, and the heart is an original part of the body.
> By transitivity, the new ventricle is original part of the body, which
> contradicts the facts.
>
very good example. Indeed, "original_part_of" is not transitive. But I
assume that
the following right idendity rules hold:
"original_part_of" o "part_of" -> "original_part_of"
as well as
"original_part_of" o "located_in" -> "original_part_of"
Stefan
>>
>> On the other hand there are good reasons not do consider offsprings as
>> part of the parent organism. They may be located in (cf. ***), but are
>> not part of. If we restrict parthood in that way (which however relies
>> on a notion of organismal unity, which may be challenged especially by
>> plant and fungi examples) then we can consider "original_part (a, b)"
>> as transitive.
>>
>> Isn't it the case that most OBO ontology users tacitly use the OBO
>> part_of relation already in this sense, as they use OBO ontologies for
>> annotating scientific data from experiments with detached mouse or
>> drosophila parts, tissue samples, etc?
>
> This deserves a longer answer as uses are varied. However, one of the common
> uses is to represent the conclusions of an experiment or series of
> experiments, as a statement about wild-type canonical organisms. E.g. that a
> specific gene product is (usually) expressed in a certain location or has a
> certain function.
>
> Mutant phenotype cases are more difficult
>
>>
>>> The alternative: consider a particular amputated mouse tail to be not an
>>> instance of MA:tail, but instead an instance of a type that stands in a
>>> transformation_of relation to MA:tail, does not seem so bad to me.
>>
>> Then we would need different classes for in situ and detached body
>> parts, which would complicate ontology construction and maintenance.
>
> Why? Why not just create a class description whenever you want to describe a
> detached body part? The situation is analagous to describing mutant
> phenotypes.
>
> Even if we needed to pre-compose these descriptions in an ontology of
> detached body parts, this could be automated. I haven't seen a need for such
> an ontology yet
>
> Cheers
> Chris
>
>>
>> Best regards
>> Stefan
>>
>> ---
>> *** Schulz S, Daumke P, Smith B, Hahn U.
>> How to distinguish parthood from location in bioontologies.
>>
>> http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=16779124
>>
>
>
--
Stefan SCHULZ (apl. Prof. Dr. med.)
Universitätsklinikum - Institut für Medizinische
Biometrie und Medizinische Informatik
Stefan-Meier-Strasse 26 D-79104 Freiburg
[home: Eschholzstr. 70 D-79115 Freiburg]
+49 (0)761 2036725, 2049089, FAX 2036711
http://purl.org/steschu
[stsc...@uni-freiburg.de], Skype: stschulz
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Obo-relations mailing list
Obo-re...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-relations
Alan Ruttenberg
> Hi Chris
>>
>> Here's another example, going back to hearts:
>>
>> Consider our surgically removed heart. Let's say the extracted heart is kept
>> alive, and we induce development of an entirely new part such as a
>> ventricle, perhaps by stem cell therapy. The new ventricle is undoubtedly an
>> original part of the heart, and the heart is an original part of the body.
>> By transitivity, the new ventricle is original part of the body, which
>> contradicts the facts.
>>
> very good example. Indeed, "original_part_of" is not transitive. But I
> assume that
> the following right idendity rules hold:
>
> "original_part_of" o "part_of" -> "original_part_of"
> as well as
> "original_part_of" o "located_in" -> "original_part_of"
Just checking this.
I read this as:
if x original_part_of y and y part_of z then x original_part_of z
If, so this seems wrong. The left ventricle valve is originally part
of Sam, and was transplanted to Joe after Sam deceased. It is now part
of Joe. This would imply that the left ventricle valve is
original_part_of Joe?
-Alan
Chris Mungall
If these are instance level relations, then there is a missing
argument for the timepoint
Using 'some' prefix for all-some type-level relations (rather than
italics):
"original_part_of_some" o "part_of_some" -> "original_part_of_some"
This seems intuitive, and I believe it follows from the definitions of
the constituent relations
Alan Ruttenberg
Yes, and part_of has a time index, but original_part_of doesn't (the
time is built into the word "original").
>
> Using 'some' prefix for all-some type-level relations (rather than italics):
> "original_part_of_some" o "part_of_some" -> "original_part_of_some"
>
> This seems intuitive, and I believe it follows from the definitions of the constituent relations
I don't see how. Would you mind restating my example so as to
hopefully remove my confusion?
Thanks,
Alan
Wacek Kusnierczyk
> On Mon, Nov 24, 2008 at 1:58 PM, Chris Mungall <c...@berkeleybop.org> wrote:
>
>
>
>> Using 'some' prefix for all-some type-level relations (rather than italics):
>> "original_part_of_some" o "part_of_some" -> "original_part_of_some"
>>
>>
suffix?
vQ
Chris Mungall
I fed some axioms into prover9 and it found a proof, see attached.
I haven't read through and checked the proof but I trust prover9. Also
the composition rules makes intuitive sense to me if I draw out a
diagram with time along the x-axis. That and I can't think of a
counter-example.
Stefan Schulz
> original_part_of o part_of -> original_part_of
> as well as
> original_part_of o located_in -> original_part_of
> If, so this seems wrong. The left ventricle valve is originally part
> of Sam, and was transplanted to Joe after Sam deceased. It is now part
> of Joe. This would imply that the left ventricle valve is
> original_part_of Joe?
But which z Sam is part_of ?
And there is no variable for Joe.
On instance vs. class level relation:
According to OBO RO there is only one ternary part_of relation: part_of (x, y, t).
[binary] part_of (x, y) iff:
for all t: part_of (x, y, t) (i)
[binary] original_part_of (x, y) iff:
for t0: part_of (x, y , t0)
given that there is no t < t0 at which x exists. (ii)
From this you can deduce that
original_part_of (x, y) AND part_of (y, z) => original_part_of (x, z)
Proof: if part_of (y, z, t) at any t and y and z exist, therefore also at t0
part_of (x, y, t0); part_of (y, z , t0); => part_of (x, z , t0) (transitivity)
x does not exist at any t < t0, according to (ii):
--
x original_part_of y means part_of at the time
2008/11/25 Chris Mungall <c...@berkeleybop.org>:
Alan Ruttenberg
> On the heart transplant example:
>
>> original_part_of o part_of -> original_part_of
>> as well as
>> original_part_of o located_in -> original_part_of
>
>> if x original_part_of y and y part_of z then x original_part_of z
>
> Yes, that is the intended meaning
>
>> If, so this seems wrong. The left ventricle valve is originally part
>> of Sam, and was transplanted to Joe after Sam deceased. It is now part
>> of Joe. This would imply that the left ventricle valve is
>> original_part_of Joe?
>
> The left ventricle valve (x) is original_part_of Sam (y).
> But which z Sam is part_of ?
> And there is no variable for Joe.
>
> On instance vs. class level relation:
> According to OBO RO there is only one ternary part_of relation: part_of (x,
> y, t).
> Abstracting from the third argument, we have different ways to form binary
> part_of subrelations, e.g
>
> [binary] part_of (x, y) iff:
> for all t: part_of (x, y, t) (i)
OK. Here is where the disconnect is. I didn't (and don't) read the
instance level part_of as true for all time. In RO the abstraction
over time happens in the class relation. Instead OWL ontologies are
typically considered SNAPs - they exist at some particular time. That
time is the implicit time argument for all ternary relations with time
in it.
To avoid confusion I think this should be stated in terms of the
binary instance level relation always_part_of relation
always_part_of(x,y) = for all t: part_of (x, y, t)
Otherwise we lose all ability to have time specific instance level
relations in OWL ontologies (think development specific anatomy).
-Alan
Alan Ruttenberg
>
>> original_part_of o part_of -> original_part_of
>> as well as
>> original_part_of o located_in -> original_part_of
>
>> if x original_part_of y and y part_of z then x original_part_of z
>
> Yes, that is the intended meaning
>
>> If, so this seems wrong. The left ventricle valve is originally part
>> of Sam, and was transplanted to Joe after Sam deceased. It is now part
>> of Joe. This would imply that the left ventricle valve is
>> original_part_of Joe?
>
> The left ventricle valve (x) is original_part_of Sam (y).
Sorry. More carefully:
At t1
Sams left ventricle valve (x) is original_part_of heart of Sam (y)
which is (original) part of of Sam (z)
Works: x is original part of z
At t2 after heart transplant
Sams left ventricle valve (x) is part_of heart of Sam (y) which is
part of of Joe (r)
Doesn't work: conclusion is that x is original part of r (which it isn't)
(with my interpretation that part_of wasn't for all t, as Sams heart
part_of Sam at t1 but part_of Joe at t2.)
-Alan