[Obo-taxonomy] Amphibian Taxonomic Ontology
Chris Mungall
obofoundry.org at midnight. It has not undergone any kind of review as
yet. For now it can be obtained from http://ontology1.srv.mst.edu/sarah/amphibian_taxonomy.obo
We need to agree on naming conventions for taxo-ontologies. ATO
includes names such as
'Bufo bufo (Linnaeus 1758)'
* there is no synonym liking this to the common name 'European toad'
* there is no need for the qualifier. I understand this is convention
in taxonomy to disambiguate different descriptions of Bufo bufo, but I
don't think we need to elevate this to the primary term label in the
ontology representation. It's not how it's handled in TTO or NCBITaxon
* there is no xref to the NCBITaxon - this is a major problem and
hindrance to data integration
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Obo-taxonomy mailing list
Obo-ta...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-taxonomy
Jonathan Rees
*any* of the numerous taxonomic databases, such as namebank or
catalogue of life. See e.g.
http://www.ubio.org/browser/details.php?namebankID=2475654 . (NCBI
taxonomy is oriented not to general taxonomy but rather only to model
organisms and genomics, so it's not necessarily the best place to link
to for this kind of a data set.)
Jonathan
Hilmar Lapp
On Aug 13, 2009, at 12:25 PM, Chris Mungall wrote:
> ATO has been registered with the obolibrary and should show up on
> obofoundry.org at midnight. It has not undergone any kind of review as
> yet. For now it can be obtained from http://ontology1.srv.mst.edu/sarah/amphibian_taxonomy.obo
>
> We need to agree on naming conventions for taxo-ontologies. ATO
> includes names such as 'Bufo bufo (Linnaeus 1758)'
> [...]
> * there is no need for the qualifier. I understand this is convention
> in taxonomy to disambiguate different descriptions of Bufo bufo, but I
> don't think we need to elevate this to the primary term label in the
> ontology representation. It's not how it's handled in TTO or NCBITaxon
It isn't but that's not necessarily a Good Thing. The qualifier
certainly will be needed if we ever want to establish exact synonym or
subsumption relationships across taxonomy ontologies as otherwise it
is rather impossible to do any machine-based matching of taxon
concepts. [1]
So I would consider the fact that it's absent from TTO a defect. I
agree it need not be in the term label - maybe have it as an exact
synonym?
> * there is no xref to the NCBITaxon - this is a major problem and
> hindrance to data integration
True. I'll add to Jonathan's post that I think this is best deferred
to a taxonomic intelligence service to do it right. Aside from uBio
there is work on-going on the Global Names Architecture [2] - I'd hope
that between those initiatives the cross-refs can be generated at some
point.
-hilmar
[1] http://www.tdwg.org/standards/117/
[2] http://gnapartnership.org/gna/wiki and http://globalnames.org
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- hlapp at duke dot edu :
===========================================================
Peter Midford
get a decent system to resolve names. However, something we should be
considering now, for all the taxonomy ontologies, is how the
standardize our use of the synonym facility OBO does provide us. We
need to define a vocabulary for synonym types: common name, valid full
name (with the author and year), and maybe try to work out the various
degrees of homonymy. It would mean we are committing to only to
names, rather than anything like a taxon concept, but I suppose with
links to standard providers (e.g., gna) it will suffice.
Peter
Peter E. Midford
Mesquite Developer
Peter....@gmail.com
Alan Ruttenberg
than having dbxrefs is there any reason ATO should just use the NCBI
taxonomy ids directly if it means to refer to the same entities?
OBI and others have been using NCBI taxonomy. Having a dbxref allows
for only painstaking mapping based data integration. If the NCBI
Taxonomy term means the same thing, then it's much simpler for all of
us if that ID is used.
-Alan
http://precedings.nature.com/search?query=mireot
Peter Midford
> At the moment, we have NCBI Taxonomy in the OBO ontologies. Rather
> than having dbxrefs is there any reason ATO should just use the NCBI
> taxonomy ids directly if it means to refer to the same entities?
>
> OBI and others have been using NCBI taxonomy. Having a dbxref allows
> for only painstaking mapping based data integration. If the NCBI
> Taxonomy term means the same thing, then it's much simpler for all of
> us if that ID is used.
>
Integrating a mix of NCBI terms with terms that will never have NCBI
ids (e.g., 300 myo fossils) into an ontology is also a headache. I
also don't know what basis we have for saying that a name in NCBI and
one in uBio or gna mean the same thing - same name, some traceable
specimen?
Peter
Peter E. Midford
Mesquite Developer
Peter....@gmail.com
------------------------------------------------------------------------------
Alan Ruttenberg
> On Aug 13, 2009, at 21:04, Alan Ruttenberg wrote:
>
>> At the moment, we have NCBI Taxonomy in the OBO ontologies. Rather
>> than having dbxrefs is there any reason ATO should just use the NCBI
>> taxonomy ids directly if it means to refer to the same entities?
>>
>> OBI and others have been using NCBI taxonomy. Having a dbxref allows
>> for only painstaking mapping based data integration. If the NCBI
>> Taxonomy term means the same thing, then it's much simpler for all of
>> us if that ID is used.
>>
>
> Integrating a mix of NCBI terms with terms that will never have NCBI ids
> (e.g., 300 myo fossils) into an ontology is also a headache.
Yes, but it's a headache that can be done once by the developer of the
ontology, instead of countless times for each user of the ontology.
> I also don't
> know what basis we have for saying that a name in NCBI and one in uBio or
> gna mean the same thing - same name, some traceable specimen?
The developer of the ontology is in a much better position to say than I am:
I see
[Term]
id: ATO:0000112
name: Rheobates palmatus
property_value: has_rank ATO:0000034 ! species
is_a: ATO:0000111 ! Rheobates
and
http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info&id=238108
On what basis, as a user of both of these ontologies, am *I* supposed
to determine if they mean the same thing?
I would suggest that if there is evidence that these two terms do not
co-refer, I should see something in the ontology that would convince
me.
-Alan
Hilmar Lapp
taxonomic synonyms. But shouldn't that be done in close coordination
with TDWG. In fact, there is already a TaxonName [1] vocabulary and
maybe what we should really do is work to ensure it fulfills these
needs.
(I don't know whether any TDWGigian is on this list, so I'm copying
Roger just in case. Roger - the TaxonName home doesn't say which
mailing list to direct any comments or questions to - is there one
that's recommended? Can you add it to the page?)
-hilmar
[1] http://rs.tdwg.org/ontology/voc/TaxonName
Hilmar Lapp
That's because you take identity of the binomial name as strong
evidence that they mean the same thing so you turn the burden of proof
around. You would need to have a reference to the same type specimen
though, or short of that, the full name with author and year. (It
doesn't help that even the same fully qualified names are not always
used consistently, for example between different floras.) So the
burden is equally on you to convince me that they *do* co-refer.
Interestingly, in your example one of the NCBI link-outs does go to
AmphibiaWeb, which I believe ATO is based on (is that true, Sarah?),
but the link is dangling (maybe AmphibiaWeb needs to update its NCBI
linkouts?). If you replace the old genus with Rheobates then you do
get a page, with links back to GenBank using only the binomial (rather
than the NCBI taxID!) as the query. That would suggest that
AmphibiaWeb considers as Rheobates palmatus whatever NCBI considers it
as. (I'm copying David Wake here in case he has any insight on this.)
There's been a lot of work on mapping taxon concepts, and I don't
think we do ourselves a favor or advance our goals by ignoring it,
which assuming that binomial name identity confers taxon concept
identity effectively constitutes. A recent paper by Franz & Peet [1]
has some nice illustrations.
Many of the use-cases for these taxonomy ontologies (such as computing
over evolutionary phenotypes) cross over into biodiversity and
conservation. Frankly, the NCBI taxonomy is not the one that ties
those fields together, so I might as well ask why GO etc don't use
ITIS or CoL IDs to support data integration. Personally I think the
way forward is taxonomic intelligence and infrastructure to map taxon
concepts largely automatically, and not arguing over which taxonomy
should rule all others or is the only one that can be used to achieve
data integration.
-hilmar
[1] Franz, N.M. and R.K. Peet. 2009. Towards a language for mapping
relationships among taxonomic concepts. Systematics and Biodiversity
(in press).
http://www.bio.unc.edu/Faculty/peet/pubs/Syst&Biodiv2009.pdf
Interesting reading regarding biodiversity application is also the
following:
Franz, N. M., R. K. Peet & A. S. Weakley. 2008. On the use of
taxonomic concepts in support of biodiversity research and taxonomy.
Symposium Proceedings, In:Wheeler, Q. D.,Ed., The New Taxonomy.
Systematics Association Special Volume (74). Taylor & Francis, Boca
Raton, FL, pp. 63–86.
http://www.bio.unc.edu/Faculty/peet/pubs/Cardiff.pdf
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- hlapp at duke dot edu :
===========================================================
------------------------------------------------------------------------------
Arlin Stoltzfus
> On Thu, Aug 13, 2009 at 10:32 PM, Peter Midford<peter....@gmail.com
> > wrote:
>> On Aug 13, 2009, at 21:04, Alan Ruttenberg wrote:
>>
>>> At the moment, we have NCBI Taxonomy in the OBO ontologies. Rather
>>> than having dbxrefs is there any reason ATO should just use the NCBI
>>> taxonomy ids directly if it means to refer to the same entities?
>>>
>>> OBI and others have been using NCBI taxonomy. Having a dbxref allows
>>> for only painstaking mapping based data integration. If the NCBI
>>> Taxonomy term means the same thing, then it's much simpler for all
>>> of
>>> us if that ID is used.
>>>
>>
>> Integrating a mix of NCBI terms with terms that will never have
>> NCBI ids
>> (e.g., 300 myo fossils) into an ontology is also a headache.
>
> Yes, but it's a headache that can be done once by the developer of the
> ontology, instead of countless times for each user of the ontology.
In other words, OBO would designate NCBI's taxonomy as its reference
taxonomy, to which all other taxonomies need to map?
From what I have heard, this isn't the outcome one would get by
consulting stakeholders in the systematics community (which is
responsible for making taxonomies).
Arlin
-------
Arlin Stoltzfus (stol...@umbi.umd.edu)
Fellow, CARB; Adj. Assoc. Prof., UMBI; Research Biologist, NIST
CARB, 9600 Gudelsky Drive, Rockville, MD
tel: 240 314 6208; web: www.molevol.org
Alan Ruttenberg
I don't agree. I used all the information available to me, which, on
the ATO side was *only* the name. If there was more information in the
ontology then I might have a chance of figuring out whether there is
or isn't a difference.
If the ontology fails to communicate what it's terms denote, then, in
my view, it fails, since the point of using a shared ontology is that
we all mean the same thing when we use terms from it. If consumers
don't have adequate information to determine, on their own, what a
term means, that can't happen. How exactly do you recommend that I
figure out what ATO means by ATO:0000112?
>
> Interestingly, in your example one of the NCBI link-outs does go to
> AmphibiaWeb, which I believe ATO is based on (is that true, Sarah?), but the
> link is dangling (maybe AmphibiaWeb needs to update its NCBI linkouts?). If
> you replace the old genus with Rheobates then you do get a page, with links
> back to GenBank using only the binomial (rather than the NCBI taxID!) as the
> query. That would suggest that AmphibiaWeb considers as Rheobates palmatus
> whatever NCBI considers it as. (I'm copying David Wake here in case he has
> any insight on this.)
>
> There's been a lot of work on mapping taxon concepts, and I don't think we
> do ourselves a favor or advance our goals by ignoring it, which assuming
> that binomial name identity confers taxon concept identity effectively
> constitutes. A recent paper by Franz & Peet [1] has some nice illustrations.
Sure. But we don't have a chance in hell of doing this unless there is
more information supplied in ATO. From my point of view there is, at
the moment, a much better chance of figuring out what NCBI means (or
if it is inconsistency) because of cross linking they have done.
> Many of the use-cases for these taxonomy ontologies (such as computing over
> evolutionary phenotypes) cross over into biodiversity and conservation.
> Frankly, the NCBI taxonomy is not the one that ties those fields together,
> so I might as well ask why GO etc don't use ITIS or CoL IDs to support data
> integration.
A perfectly reasonable question. Look, OBO is a community and we try
to coordinate as best possible, given what people are willing to
contribute. We desperately need a standard way of identifying species
and the only candidate in the current set of contributions is NCBI
Taxonomy. So a number of groups have said - ok let's use that until
something else better comes out.
Now, if someone contributed an OBO/OWL version of an alternative and
this was clearly a better resource, then I'd use it and lobby others
to use it.
But that's not what we have here. We have a resource that, at least to
a certain extent, is *inferior* to the existing resource, and gives
new identifiers to terms that most likely *do* co-refer. That's moving
in the wrong direction (no offense to the submitters - I'm trying to
emphasize a technical point here).
> Personally I think the way forward is taxonomic intelligence
> and infrastructure to map taxon concepts largely automatically, and not
> arguing over which taxonomy should rule all others or is the only one that
> can be used to achieve data integration.
Sounds idyllic. If it could only work. All experience I have says that
unless there is tighter coordination this won't succeed.
-Alan
Alan Ruttenberg
> On Aug 13, 2009, at 10:43 PM, Alan Ruttenberg wrote:
>
>> On Thu, Aug 13, 2009 at 10:32 PM, Peter Midford<peter....@gmail.com
>> > wrote:
>>> On Aug 13, 2009, at 21:04, Alan Ruttenberg wrote:
>>>
>>>> At the moment, we have NCBI Taxonomy in the OBO ontologies. Rather
>>>> than having dbxrefs is there any reason ATO should just use the NCBI
>>>> taxonomy ids directly if it means to refer to the same entities?
>>>>
>>>> OBI and others have been using NCBI taxonomy. Having a dbxref allows
>>>> for only painstaking mapping based data integration. If the NCBI
>>>> Taxonomy term means the same thing, then it's much simpler for all
>>>> of
>>>> us if that ID is used.
>>>>
>>>
>>> Integrating a mix of NCBI terms with terms that will never have
>>> NCBI ids
>>> (e.g., 300 myo fossils) into an ontology is also a headache.
>>
>> Yes, but it's a headache that can be done once by the developer of the
>> ontology, instead of countless times for each user of the ontology.
>
> In other words, OBO would designate NCBI's taxonomy as its reference
> taxonomy, to which all other taxonomies need to map?
Until there is something better. And I'm not talking about "mapping".
I'm talking about not inventing new words when the old ones work. One
should never dumb down by mapping to a term which doesn't mean the
same thing.
> From what I have heard, this isn't the outcome one would get by
> consulting stakeholders in the systematics community (which is
> responsible for making taxonomies).
I would *love* there to be more participation from that community. But
that participation has to result in ontologies that can be reliably
used. That means including enough information so that one can figure
out what a term means. One shouldn't forget that the first priority of
an ontology is clear communication.
-Alan
Nico Franz
I noticed that someone cited a couple of references in which I was
involved (thanks! at the time I felt that was a little progress...). I
should mention that Dave Thau (http://wwwcsif.cs.ucdavis.edu/~thau/) is
taking this issue - taxonomy, mapping, ontology - into a new and very
interesting direction.
I remain interested in contributing to a more formal representation
and semantic linking of multiple taxonomies, although given that
systematists are trying to constantly improve classifications of nature,
I sense that there are some very special problems with representing
systematic progress that way (i.e. though an ontology). In some
taxonomic groups the only constant is change, and this is different from
e.g. a medical or gene ontology where the fixation of terms, meanings,
and relations is more a matter of convention, perhaps, and isn't
constantly overturned by the the "discovery" (inference) of new homology
statements, phylogenetic relationships, and species hypotheses, etc.
Even though methodologically systematics is a fairly well established
field, in terms of semantic content the stated goal of most taxonomic
papers is to actually change the established paradigm, scratch the
previous representations of causal relationships in nature, at propose
new ones. I am overstating but hopefully not by much. I do feel that the
uniqueness of systematics vis-a-vis representability with computer
ontologies has not been adequately explored. There may be some very
special problems that are fundamentally different from what ontology
textbooks tend to emphasize. I allude to this towards the very end (pp.
10-12) of the following "chapter":
http://academic.uprm.edu/~franz/publications/LetterLinnaeus.pdf
Regards,
Nico
Nico M. Franz
Department of Biology
University of Puerto Rico
Call Box 9000
Mayagüez, PR 00681-9000
Phone: (787) 832-4040, ext. 3005
Fax: (787) 834-3673
E-mail: nico....@upr.edu
Website: http://academic.uprm.edu/~franz/
Nico Franz
I noticed that someone cited a couple of references in which I was
involved. I should mention that Dave Thau
(http://wwwcsif.cs.ucdavis.edu/~thau/) is taking this issue - taxonomy,
mapping, ontology - into a new and very interesting direction. I have
limited knowledge of ontologies and computer science in general, but
remain interested in contributing to a more formal representation and
semantic linking of multiple taxonomies.
Regards,
Nico
Nico M. Franz
Department of Biology
University of Puerto Rico
Call Box 9000
Mayagüez, PR 00681-9000
Phone: (787) 832-4040, ext. 3005
Fax: (787) 834-3673
E-mail: nico....@upr.edu
Website: http://academic.uprm.edu/~franz/
obo-taxono...@lists.sourceforge.net wrote:
> Send Obo-taxonomy mailing list submissions to
> obo-ta...@lists.sourceforge.net
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://lists.sourceforge.net/lists/listinfo/obo-taxonomy
> or, via email, send a message with subject or body 'help' to
> obo-taxono...@lists.sourceforge.net
>
> You can reach the person managing the list at
> obo-taxon...@lists.sourceforge.net
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Obo-taxonomy digest..."
>
>
> Today's Topics:
>
> 1. ncbi_taxonomy.obo? (Peter Midford)
> 2. Re: Amphibian Taxonomic Ontology (Hilmar Lapp)
> 3. Re: Amphibian Taxonomic Ontology (Hilmar Lapp)
> 4. Re: Amphibian Taxonomic Ontology (Arlin Stoltzfus)
> 5. Re: Amphibian Taxonomic Ontology (Alan Ruttenberg)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Fri, 14 Aug 2009 06:11:22 -0500
> From: Peter Midford <peter....@gmail.com>
> Subject: [Obo-taxonomy] ncbi_taxonomy.obo?
> To: Chris Mungall <c...@berkeleybop.org>
> Cc: obo-ta...@lists.sourceforge.net
> Message-ID: <1A46E861-340E-4AE7...@gmail.com>
> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
>
> Chris,
> I notice the obo file for the ncbi taxonomy has
> disappeared from the foundry. Is this an oversight or does it
> represent something else?
>
> Peter
>
> Peter E. Midford
> Phenoscape Ontology Curator
> Peter....@gmail.com
>
>
>
>
>
>
>
> ------------------------------
>
> Message: 2
> Date: Fri, 14 Aug 2009 10:13:53 -0400
> From: Hilmar Lapp <hl...@duke.edu>
> Subject: Re: [Obo-taxonomy] Amphibian Taxonomic Ontology
> To: Peter Midford <peter....@gmail.com>
> Cc: Sarah Havens <sbh...@mst.edu>, Chris Mungall
> <c...@berkeleybop.org>, obo-ta...@lists.sourceforge.net
> Message-ID: <79AA222D-0C17-471A...@duke.edu>
> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
>
> I agree that there should to be a commonly agreed upon vocabulary for
> taxonomic synonyms. But shouldn't that be done in close coordination
> with TDWG. In fact, there is already a TaxonName [1] vocabulary and
> maybe what we should really do is work to ensure it fulfills these
> needs.
>
> (I don't know whether any TDWGigian is on this list, so I'm copying
> Roger just in case. Roger - the TaxonName home doesn't say which
> mailing list to direct any comments or questions to - is there one
> that's recommended? Can you add it to the page?)
>
> -hilmar
>
> [1] http://rs.tdwg.org/ontology/voc/TaxonName
>
> On Aug 13, 2009, at 9:50 PM, Peter Midford wrote:
>
>
>> I've been following this discussion and I hope that we do eventually
>> get a decent system to resolve names. However, something we should
>> be considering now, for all the taxonomy ontologies, is how the
>> standardize our use of the synonym facility OBO does provide us. We
>> need to define a vocabulary for synonym types: common name, valid
>> full name (with the author and year), and maybe try to work out the
>> various degrees of homonymy. It would mean we are committing to
>> only to names, rather than anything like a taxon concept, but I
>> suppose with links to standard providers (e.g., gna) it will suffice.
>>
>>
>>
>> Peter
>>
>> On Aug 13, 2009, at 20:16, Hilmar Lapp wrote:
>>
>>
>>> On Aug 13, 2009, at 12:25 PM, Chris Mungall wrote:
>>>
>>>
>>>> ATO has been registered with the obolibrary and should show up on
>>>> obofoundry.org at midnight. It has not undergone any kind of
>>>> review as
>>>> yet. For now it can be obtained from http://ontology1.srv.mst.edu/sarah/amphibian_taxonomy.obo
>>>>
>>>> We need to agree on naming conventions for taxo-ontologies. ATO
>>>> includes names such as 'Bufo bufo (Linnaeus 1758)'
>>>> [...]
>>>> * there is no need for the qualifier. I understand this is
>>>> convention
>>>> in taxonomy to disambiguate different descriptions of Bufo bufo,
>>>> but I
>>>> don't think we need to elevate this to the primary term label in the
>>>> ontology representation. It's not how it's handled in TTO or
>>>> NCBITaxon
>>>>
>>> It isn't but that's not necessarily a Good Thing. The qualifier
>>> certainly will be needed if we ever want to establish exact synonym
>>> or
>>> subsumption relationships across taxonomy ontologies as otherwise it
>>> is rather impossible to do any machine-based matching of taxon
>>> concepts. [1]
>>>
>>> So I would consider the fact that it's absent from TTO a defect. I
>>> agree it need not be in the term label - maybe have it as an exact
>>> synonym?
>>>
>>>
>>>> * there is no xref to the NCBITaxon - this is a major problem and
>>>> hindrance to data integration
>>>>
>>> True. I'll add to Jonathan's post that I think this is best deferred
>>> to a taxonomic intelligence service to do it right. Aside from uBio
>>> there is work on-going on the Global Names Architecture [2] - I'd
>>> hope
>>> that between those initiatives the cross-refs can be generated at
>>> some
>>> point.
>>>
>>> -hilmar
>>>
>>> [1] http://www.tdwg.org/standards/117/
>>> [2] http://gnapartnership.org/gna/wiki and http://globalnames.org
>>>
>>> --
>>> ===========================================================
>>> : Hilmar Lapp -:- Durham, NC -:- hlapp at duke dot edu :
>>> ===========================================================
>>>
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Let Crystal Reports handle the reporting - Free Crystal Reports
>>> 2008 30-Day
>>> trial. Simplify your report design, integration and deployment -
>>> and focus on
>>> what you do best, core application coding. Discover what's new with
>>> Crystal Reports now. http://p.sf.net/sfu/bobj-july
>>> _______________________________________________
>>> Obo-taxonomy mailing list
>>> Obo-ta...@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/obo-taxonomy
>>>
>> Peter E. Midford
>> Mesquite Developer
>> Peter....@gmail.com
>>
>>
>>
>>
>>
>
>
Peter Midford
Thanks, Thau's page does have some interesting stuff.
Peter