Re: Taxonomic ranks
Alan Ruttenberg
On Apr 28, 2008, at 10:18 PM, Peter Midford wrote:
> Alan,
> Thanks for the discussion of metaclasses. I had assumed
> they were sort of off limits in order to keep OBO approximately
> equivalent to OWL-DL.
Yes, that is a problem. However, if you use annotation properties
where you have class/class links then you can make the links,
although inferences not made. On the other hand, is it clear what
inferences you want to be able to make? In OWL 2 you will be able to
use punning to get a bit more expressivity. In OWL 1 you could manage
the punning manually (by creating a differently named instance for
each class that needs to participate in an instance level
relationship. Overall I would take as a given that OWL won't infer
everything you want, without inferring also things that you don't want.
> The bottom line for us at the present is a desire to see species
> and higher taxa as individuals so we can meaningfully link them,
> when appropriate, to clades on phylogenetic trees. There would
> still be an ontology of taxonomic ranks, but the taxonomy itself
> should be, or be viewable as, an instance graph. Then particular
> species, genera etc. would be instances of their level term, which
> in turn would be subclasses of a class of clades. Part_of seems to
> us to be an appropriate relation between nested portions of
> phylogenetic trees.
Here we disagree, though we agree on the undesired inference, which
gives us something to build on.
> Getting from a clade to an individual organism is trickier, though
> I have found it helpful to use lineages as an intermediate between
> species and individuals (if you are familiar with coalescent
> models, I visualize it like a gene tree within a species tree).
> Jim Balhoff and I agree that the transitive inference that my arm
> is part_of Homo sapiens is a problem.
>
> Being able to map taxonomy terms to nodes/clades on phylogenetic
> trees has been mentioned multiple times in response to Phenoscape
> presentations, so we are taking it seriously.
>
> Jim has illustrated our idea in a nice diagram:
> https://phenoscape.org/wg/phenoscape/images/thumb/7/7c/
> TaxonomicRankDiagram.png/750px-TaxonomicRankDiagram.png
See my attached drawing. In it I've layed out what something I think
is closer, along with one inference that should be made for the
sub_rank_of relation. I'm not entirely thrilled with it, as I like to
ground out all class/class relations as expressions on instances, and
I don't know how to do that here (how do I express the relationship
between "species" and "rank" in terms of organisms?)
There are three levels of classes: class (Danio) metaclass (genus)
and meta-meta-class (rank). At each level the class at the previous
level is considered an instance at the next.
Unstated are what the differentia are between the subclasses of
taxon, though I alluded to what I thought the nature of them were in
my previous message.
Perhaps there is something here to build on. If you don't mind, I'm
ccing the OBI Biomaterial group as they will be interested in the
discussion.
Alan Ruttenberg
On Apr 29, 2008, at 9:40 AM, Hilmar Lapp wrote:
> Hi Alan,
>
> in my understanding is-a implies inheritance of properties
> (subsumption): if A is-a B, then every instance of A also is an
> instance of B, and has all properties of B (plus those of A).
Every instance of A necessarily has any properties that any every
instance of B necessarily has. One has to be careful about mixing
together the "properties of the class" versus the properties of the
instances of the class.
> This is in fundamental disagreement with (my understanding of) the
> tenets of evolution, so I don't see how Danio is-a Cypriniformes
> can hold.
Anything that is necessarily true of any Cypriniformes is true of any
Danio.
> However, evolutionarily, all organisms that are a member of (part
> of?) the Danio genus have a common ancestor that was a member of
> the Cypriniformes, even if they have very different properties now.
That there was a common ancestor *is* a property.
> Much like all humans and chimps have a common ancestor that was
> also a member of the primates, although humans (in all likelihood)
> don't have all the properties that our common ancestor with chimps
> had.
>
> Which brings me to the other issue I have with is-a: which
> properties are inherited and which are modified, or in other words,
> which property values are ancestral and which are derived, are a
> matter of scientific debate and represent hypotheses; they should
> (in my opinion) not be encoded in an ontology.
Indeed, that is the question to ask, and only correct answers ought
to be encoded in the ontology.
-Alan
> -hilmar
>> <Taxonomic.pdf>
>>>
>>> I'll let you know when the accompanying page is ready.
>>>
>>> Peter
>>>
>>>
>>> Peter E. Midford
>>> Phenoscape Ontology Curator
>>> Peter....@gmail.com
>>>
>>>
>>>
>>>
>>
>> _______________________________________________
>> Phenoscape mailing list
>> Pheno...@nescent.org
>> https://lists.nescent.org/mailman/listinfo/phenoscape
>
> --
> ===========================================================
> : Hilmar Lapp -:- Durham, NC -:- hlapp at duke dot edu :
> ===========================================================
>
>
>
>
Barry Smith
>Hi Alan,
>
>On Apr 29, 2008, at 11:03 AM, Alan Ruttenberg wrote:
>
>>On Apr 29, 2008, at 9:40 AM, Hilmar Lapp wrote:
>>
>>>This is in fundamental disagreement with (my understanding of) the
>>>tenets of evolution, so I don't see how Danio is-a Cypriniformes
>>>can hold.
>>
>>Anything that is necessarily true of any Cypriniformes is true of
>>any Danio.
>
>properties of so-called "instances" of Cypriniformes, we also want to
>say things about Cypriniformes itself, and Danio itself, which are
>real unique things in history (e.g. date of origin, extinction,
>geographic range, morphological diversity, etc.).
How about the view, defended most articulately by Ghiselin
http://dannyreviews.com/h/Metaphysics_Species.html,
to the effect that the species is the mereological sum of instances.
This is a real unique (scattered) thing in history, with date of
origin, etc., etc.
> My impression is
>that it doesn't make sense to attach particular property values to
>ontology classes (if we do treat these taxa as ontology classes).
Ontologies, sadly, are none too clear about what they mean by
'class'; but they seem NOT to be entities which have a date of
origin, extinction, etc., thus not to be entities subject to
biological change. I think we need to look elsewhere, e.g. to Ghiselin.
In Ghiselin's terms to say
Danio is_a Cypriniformes
is, in first approximation to say that the mereological sum of all
instances of Danio is a part of the mereological sum of all instances
of Cypriniformes AND anything that is necessarily true of any
Cypriniformes is true of any Danio.
BS
>I am very glad we're having this discussion - I like the model we came
>up with, but it sounds like there are other possible alternatives to
>the default "is_a" construction.
>
>Thanks,
>Jim
>
>____________________________________________
>James P. Balhoff, Ph.D.
>National Evolutionary Synthesis Center
>2024 West Main St., Suite A200
>Durham, NC 27705
>USA
>
>
>
Alan Ruttenberg
How about the view, defended most articulately by Ghiselin
http://dannyreviews.com/h/Metaphysics_Species.html,
to the effect that the species is the mereological sum of instances.
This is a real unique (scattered) thing in history, with date of
origin, etc., etc.
My impression is
that it doesn't make sense to attach particular property values to
ontology classes (if we do treat these taxa as ontology classes).
Ontologies, sadly, are none too clear about what they mean by
'class'; but they seem NOT to be entities which have a date of
origin, extinction, etc., thus not to be entities subject to
biological change. I think we need to look elsewhere, e.g. to Ghiselin.
In Ghiselin's terms to say
Danio is_a Cypriniformes
is, in first approximation to say that the mereological sum of all
instances of Danio is a part of the mereological sum of all instances
of Cypriniformes AND anything that is necessarily true of any
Cypriniformes is true of any Danio.
Chris Mungall
The alternative relegates the term "species" to denote a class of
classes, which would be unacceptable if the term "species"
corresponded to a unit in scientific theories - this is a somewhat
garbled paraphrasing of some of the arguments put forward in the
stanford encyclopedia article Jim linked to.
There are also Jim's arguments (my paraphrasing / interpretation)
that qualities such as diversity inhere in individuals and not
classes. This seems a fairly solid practical reason (I can think of
non evo use cases where it would be useful to hang these kinds of
properties off of classes, but this just doesn't work well with the
formalisms we have at the moment)
My personal view is that it is at different times useful to talk of
HomoSapiens the class, HomoSapiens the continuant-population and
HomoSapiens the occurrent-lineage. Punning on a single identifier
would be a bad idea. Whilst it may seem unwise to pollute identifier
space further (how many IDs and URIs do we have for the various
senses of "P53" these days?) I think it may be to our advantage here,
as it (AFAIK) corresponds to a sort of epistemological and workflow
distinction in taxonomy/phylogeny: as hypotheses about lineages
become more established they become enshrined in taxonomies
(roughly..). I'm guessing that for most OBI purposes they want a
class C Elegans from an established source like NCBI.
Of course OBI could be used to describe investigations into evolution
itself: here's another case where lineage instances and the like are
more usable that classes.
> -Alan