[Obo-taxonomy] taxonomy ranks ontology?
Erick Antezana
is there an ontology "out there" capturing the taxonomy ranks (e.g.
species, genus) and their relationships?
I found the "Taxonomic rank vocabulary" @ BioPortal
(http://purl.bioontology.org/ontology/TAXRANK); however, it is a sort
of flat collection of terms...
Likewise, the "NCBI organismal classification"
(http://www.obofoundry.org/cgi-bin/detail.cgi?id=ncbi_taxonomy)
captures the ranks but no "hierarchy" among them...
I am looking for something that would place for instance genus below
family and above species...
thanks,
Erick
------------------------------------------------------------------------------
Create and publish websites with WebMatrix
Use the most popular FREE web apps or write code yourself;
WebMatrix provides all the features you need to develop and
publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
_______________________________________________
Obo-taxonomy mailing list
Obo-ta...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-taxonomy
Peter Midford
Why do you think that the ranks themselves should be hierarchical? Saying that a particular species is_a particular genus (or is part_of a genus) is different from saying the ranks themselves are related in that way. Both TAXRANK and the rank terms in the NCBI taxonomy are intended to be used as annotation properties or tags, it's not clear what reasoning you would expect to do with them.
Peter
Peter E. Midford
Phenoscape Ontology Curator
Peter....@gmail.com
Peter Midford
I think the question is about taxonomic ranks, not taxa that may be assigned those ranks. The assertions are of the form:
taxon F has rank Family
taxon G has rank Genus
and what sort of assertions we would make relating the terms Family and Genus.
Peter
On Mar 31, 2011, at 8:14, Jonathan Rees wrote:
> Peter,
>
> Naively one would say something like:
>
> individual A is a member of genus G
> individual A is a member of family F
> individual B is a member of genus G
> entail
> individual B is a member of family F
>
> and I think this is what the original question assumed.
>
> Are you saying that there are situations when this is not true? What about
>
> individual A is a member of genus G
> individual A is a member of family F1
> no member of family F1 is a member of family F2 or vice versa
> individual B is a member of genus G
> individual B is a member of family F2
>
> Are there any conditions (assignments of G, F1, F2) under which this is true?
>
> Maybe you don't want to say that genera or families have members, or
> that they do not reflect any theory about nature but rather are
> statements about taxon assignments that people have made. Then what
> if you replace the "members of" above with "may be correctly annotated
> with":
>
> individual A can be annotated with genus G
> individual A can be annotated with family F
> individual B can be annotated with genus G
> imply
> individual B can be annotated with family F
>
> What's a situation where this not hold?
>
> The general question is, under what circumstances would I be making a
> mistake by saying that something is or is not annotated with a
> particular rank. Is the correctness criterion based on whether any of
> a particular set of authorities (NCBI, TDWG) has asserted the rank on
> something? Or maybe a particular different authority connected each
> taxon [or taxon concept], i.e. if Joe Expert did not say that x is
> annotated with y, then it's not correct for anyone to say it? If one
> is going to use these terms one has to be able to tell whether one is
> using them correctly.
>
> I don't mean to be critical but rather to simulate the questions a
> naive observer will bring to this vocabulary, so that they can be
> answered clearly. You have probably documented your design in the
> earlier email threads and it's just taking me this long to catch up
> with you. But it would be nice if the method for using these rank
> terms were laid out on the documentation page
> https://www.phenoscape.org/wiki/Taxonomic_Rank_Vocabulary in case
> someone else comes along later with the same questions, which I think
> is likely.
>
> I don't think the idea of correctness disappears just because you use
> annotation properties or tags. The reasoning is then done by people,
> not by algorithms, and the correctness criteria may become more
> nuanced, but that doesn't mean it's not reasoning.
>
> Best
> Jonathan
Peter Midford
You're quite right about different fields (authorities) having different 'stacks' of terms and that's another (practical engineering) reason for not imposing an ordering on a vocabulary of rank terms that attempts to be inclusive.
Peter
On Mar 31, 2011, at 7:46, Chris Taylor wrote:
> Is this more about cultivars, tribes, races, strains and all
> those rather-more-poorly-defined terms (some of which can also
> take the super/sub prefix just to make it even more fun --
> certainly I've seen 'supertribe' used -- may have been
> dog-related I forget). Each discipline seems to have their own
> stack of these and I'm never sure exactly how they stand wrt
> each other...
>
> Tbh though the 'main' ones are just a linear thing no (KPCOFGS),
> just leaving the argument to be about the relations (is there
> something in RO like 'contains' or the inverse)?
>
> What is it you want to tag anyway -- literature-related?
>
> C.
>> Obo-discuss mailing list
>> Obo-d...@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/obo-discuss
>>
>>
>> -----
>> No virus found in this message.
>> Checked by AVG - www.avg.com
>> Version: 10.0.1209 / Virus Database: 1500/3542 - Release Date: 03/31/11
>>
>>
>
> --
> _______________________
> chris....@ebi.ac.uk
> http://mibbi.org/
>
>
> ------------------------------------------------------------------------------
> Create and publish websites with WebMatrix
> Use the most popular FREE web apps or write code yourself;
> WebMatrix provides all the features you need to develop and
> publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
> _______________________________________________
> Obo-discuss mailing list
> Obo-d...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obo-discuss
Jonathan Rees
Best
Jonathan
Erick Antezana
I actually avoided in my original email the word "hierarchical" to
implicitly ask for a "more general" relationship (other that is_a)
among ranks. I agree that such a categorisation could be a source of
long discussions among taxonomists ... but I thought that there must
be something for the 'main' ones K-P-C-O-F-G-S (as Chris pointed out)
which could extrapolated to the exotic ones (e.g. subtribe)...
Let me exemplify one of the applications in my ontology, I have to tag
a set of terms (particular species and varietes such as 'Vitis
vinifera' and 'Cucumis sativus var. sativus') as belonging to a subset
of (internal) interest. This tagging should take into account (affect)
not only the term itself but the "parent" terms as well as long as
they belong to a given rank (plus the taxa from that rank down to the
original term). So, if I pick up the 'Cucumis sativus var. sativus'
taxon and if I want to tag it as a member of a subset -- named READY
for instance, and if I also want to tag its "parents" till the family
rank, then all the following taxa will be tagged (ideally
automatically):
'Cucumis sativus var. sativus' (varietas),
'Cucumis sativus' (species),
'Cucumis' (genus),
'Cucurbitaceae' (family)
so, I would need a mechanism to *walk through* the ranks... I am not
planing to perform such tagging manually (I have a long list of
taxa...)
As a second "more interesting" application, we would like to perform
this type of "reasoning":
individual a "of species" S that "belongs_to" genus G
individual a "of species" S that "belongs_to" family F
individual b "of species" S "belongs_to" genus G
so :
individual B ("of species" S) "belongs_to" family F
cheers,
Erick
Jonathan Rees
> Jonathan,
> I think the question is about taxonomic ranks, not taxa that may be assigned those ranks. The assertions are of the form:
> taxon F has rank Family
> taxon G has rank Genus
>
> and what sort of assertions we would make relating the terms Family and Genus.
Suppose someone wrote the statement 'taxon F has rank Family'. How
would I (say, in the role of reviewer or copy editor) check to make
sure that it is correct?
Suppose I tried to publish a taxonomy in which I made the above
statements, but where the members of taxon F were a proper subclass of
the members of taxon G. Would that be a mistake? If I were reviewing
such a taxonomy, would I be entitled to send it back for revision?
Jonathan
Pierre Grenon
<erick.a...@gmail.com> wrote:
> Hi,
>
> is there an ontology "out there" capturing the taxonomy ranks (e.g.
> species, genus) and their relationships?
>
> I am looking for something that would place for instance genus below
> family and above species...
this may be of limited help, as I'm not sure how to best go about
extracting it from the whole ontology, but it's fairly well treated in
Cyc
See, for starter:
http://sw.opencyc.org/concept/Mx4rvVji6JwpEbGdrcN5Y29ycA
http://www.cyc.com/cycdoc/vocab/biology-vocab.html#BiologicalTaxonType
best
p
Adam M. Goldstein
This doesn't help Eric with his original question---unfortunately I think he is going to have to do some ontological engineering---but I think the question asked earlier in the thread, What are you categorizing, would be a useful one to answer. I would wonder as well whether the source of the info w/ the objects being categorized uses some implicit taxonomic method, and then that could be used to inform decisions about how they will be represented in your ontology.
------------------
Adam M. Goldstein PhD, MSLIS
--
z_calif...@shiftingbalance.org
http://www.shiftingbalance.org
http://www.twitter.com/shiftingbalance
--
http://www.itis.gov/servlet/SingleRpt/SingleRpt?search_topic=TSN&search_value=180621
--
(914) 637-2717 (msg)
--
Dept of Philosophy
Iona College
715 North Avenue
New Rochelle NY 10801
http://www.iona.edu/faculty/agoldstein
On Mar 31, 2011, at 7:46, Chris Taylor <chris....@ebi.ac.uk> wrote:
> Is this more about cultivars, tribes, races, strains and all
> those rather-more-poorly-defined terms (some of which can also
> take the super/sub prefix just to make it even more fun --
> certainly I've seen 'supertribe' used -- may have been
> dog-related I forget). Each discipline seems to have their own
> stack of these and I'm never sure exactly how they stand wrt
> each other...
>
> Tbh though the 'main' ones are just a linear thing no (KPCOFGS),
> just leaving the argument to be about the relations (is there
> something in RO like 'contains' or the inverse)?
>
> What is it you want to tag anyway -- literature-related?
>
> C.
>
>
> On 31/03/2011 12:30, Peter Midford wrote:
>> Obo-discuss mailing list
>> Obo-d...@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/obo-discuss
>>
>>
>> -----
>> No virus found in this message.
>> Checked by AVG - www.avg.com
>> Version: 10.0.1209 / Virus Database: 1500/3542 - Release Date: 03/31/11
>>
>>
>
> --
> _______________________
> chris....@ebi.ac.uk
> http://mibbi.org/
>
>
> ------------------------------------------------------------------------------
> Create and publish websites with WebMatrix
> Use the most popular FREE web apps or write code yourself;
> WebMatrix provides all the features you need to develop and
> publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
> _______________________________________________
> Obo-discuss mailing list
> Obo-d...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obo-discuss
------------------------------------------------------------------------------
Chris Taylor
those rather-more-poorly-defined terms (some of which can also
take the super/sub prefix just to make it even more fun --
certainly I've seen 'supertribe' used -- may have been
dog-related I forget). Each discipline seems to have their own
stack of these and I'm never sure exactly how they stand wrt
each other...
Tbh though the 'main' ones are just a linear thing no (KPCOFGS),
just leaving the argument to be about the relations (is there
something in RO like 'contains' or the inverse)?
What is it you want to tag anyway -- literature-related?
C.
On 31/03/2011 12:30, Peter Midford wrote:
> Obo-discuss mailing list
> Obo-d...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obo-discuss
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 10.0.1209 / Virus Database: 1500/3542 - Release Date: 03/31/11
>
>
--
_______________________
chris....@ebi.ac.uk
http://mibbi.org/
Kevin Richards
So we have properties of a rank that include:
RankName
KnownAbbreviations
SortOrder
IncludeInFullName - this is used to determine if the rank name should be included in the textual version of a taxon name (eg. var.)
Different ranks also imply a different approach to any "processing" pf a taxon. E.g. taxa that are "below" Genus are defined using rules of nomenclature and have a nomenclatural "parent". Higher level ranks are primarily "taxon concepts" and will vary a lot depending whose subscription you follow.
Saying this, we don’t use a standard ontology or vocabulary for our taxon ranks, but have built up the list ourselves through the biosystematics work we do.
So not sure this helps much.
TDWG (Taxonomic Databases Working Group, www.tdwg.org) is probably a good place to look (and push any improvements to such an ontology).
Kevin Richards
For interest I have included the list of taxon ranks and sort order we use in our Plant Names Database:
RankName Abbreviation KnownAbbreviations RankSort IncludeInFullName
none none @none@unranked@ 200 0
Domain Dom. @Dom.@ 300 0
Empire Emp. @Emp.@ 300 0
kingdom kingdom @kingdom@reg.@ 400 0
subkingdom subkingdom @subkingdom@subreg@ 600 0
superphylum superphylum @superphylum@superphyl.@ 700 0
Division Division @Division@div.@ 800 0
phylum phylum @phylum@phyl.@ 800 0
subphylum subphylum @subphylum@subphyl.@ 1000 0
subdivision subdiv. @subdiv.@subdivision@ 1000 0
group group @group@ 1000 0
grex grex @grex@ 1000 0
superclass superclass @superclass@supercl.@ 1100 0
class class @class@cl.@ 1200 0
subclass subclass @subclass@subcl.@ 1300 0
infraclass infraclass @infraclass@infracl.@ 1400 0
superorder superorder @superorder@superord.@ 1500 0
order order @order@ord.@ 1600 0
infraorder infraorder @infraorder@infraord.@ 1800 0
suborder suborder @suborder@subord.@ 1800 0
superfamily superfamily @superfamily@superfam.@ 1900 0
family family @family@fam.@ 2000 0
subfamily subfamily @subfamily@subfam.@ 2200 0
[infrafam.unranked] [infrafam.unranked] @[infrafam.unranked]@ 2200 0
supertribe supertribe @supertribe@supertrib.@ 2300 0
tribe tribe @tribe@trib.@ 2400 0
supersubtribe supersubtribe @supersubtrib.@ 2500 0
subtribe subtribe @subtribe@subtrib.@ 2600 0
anamorph anamorph @anamorph@ 2800 0
genus genus @genus@gen.@Genus@ 3000 0
tax. infragen. t.infragen. @t.infragen.@taxinfragen@[infragen.unranked]@tax. infragen.@ 3100 1
microgen. microgen. @microgen.@ 3200 0
apomict apomict @apomict@ap.@ 3200 0
subgenus subg. @subgenus@subgen.@subg.@ 3200 1
section sect. @section@sect.@ 3400 1
nothosection nothosect. @nothosect.@ 3400 0
subsection subsect. @subsection@subsect.@ 3600 1
series series @series@ser.@ 3800 1
subseries subseries @subseries@subser.@ 4000 1
aggr. aggr. @aggr.@ 4100 0
coll. sp. coll. sp. @coll. sp.@ 4100 0
species species @species@sp.@spec.@ 4200 0
tax. infrasp. t.infrasp. @taxin...@t.infrasp.@tax. infrasp.@ 4300 1
[infrasp.unranked] [infrasp.unranked] @[infrasp.unranked]@ 4300 0
nothosubspecies nothosub. @nothosubsp.@ 4400 0
subspecies subsp. @subsp@subsp.@ssp.@subspecies@ 4400 1
nothovariety nothovar. @nothovar.@ 4600 0
variety var. @var.@var@variety@ 4600 1
convar convar @convar.@ 4600 0
race race @race@ 4600 0
pars. pars. @pars.@ 4600 0
nm. nm. @nm.@ 4600 0
mut. mut. @mut.@ 4600 0
subvariety subvar. @subvar.@subvariety@ 4800 1
forma f. @f.@fm.@forma@fo.@ 5000 1
subforma subforma @subforma@subf.@subfo.@ 5200 1
serovar serovar @serovar@ 5400 0
pathovar pv. @pv.@pathovar.@ 5400 0
forma specialis f.sp. @f.sp.@f. sp.@ 5400 0
phagovar phagovar @phagovar@ 5400 0
biovar biovar @biovar@biovar.@ 5400 0
proles prol. @proles@prol.@ 5400 0
e e @e@ε@ 5600 1
α α @α@[alpha]@ 5600 0
γ γ @γ@ 5600 0
δ δ @δ@ 5600 0
ß ß @ß@[beta]@[beta].@beta@ß@ 5600 0
cultivar cv @cv@cult.@ 5800 0
graft hybrid graft hybrid @graft hybrid@ 5800 0
hybrid formula hybrid formula @hybrid formula@ 5800 0
intergen hybrid intergen hybrid @intergen hybrid@ 5800 0
intragen hybrid intragen hybrid @intragen hybrid@ 5800 0
lus lus @lusus@lus.@ 5800 0
Warning: This electronic message together with any attachments is confidential. If you receive it in error: (i) you must not read, use, disclose, copy or retain it; (ii) please contact the sender immediately by reply email and then delete the emails.
The views expressed in this email may not be those of Landcare Research New Zealand Limited. http://www.landcareresearch.co.nz
Richard Pyle
implementations (including ZooBank), and is the basic architecture for how
Ranks will be managed in the Global Names Architecture data model. I would
be interested to hear what others think about how much (or little) ranks
should be modeled. I think they are definitely more than a simple
enumeration of terms that serve as attributes of names (or, more precisely,
attributes of taxon name usage instances). However, they may not rise to
the level of requiring a complete ontological infrastructure. Is there any
domain in-between these two extremes?
Aloha,
Rich
> _______________________________________________
> Obo-taxonomy mailing list
> Obo-ta...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obo-taxonomy
>
> Please consider the environment before printing this email
> Warning: This electronic message together with any attachments is
> confidential. If you receive it in error: (i) you must not read, use,
disclose,
> copy or retain it; (ii) please contact the sender immediately by reply
email
> and then delete the emails.
> The views expressed in this email may not be those of Landcare Research
> New Zealand Limited. http://www.landcareresearch.co.nz
> --------------------------------------------------------------------------
----
Peter Midford
How do you deal with differences in usage between authorities (e.g., the respective bodies for Botany and Zoology)?
Specifying a sort order field suggests that inserting intermediate level rank terms causes field values to change, which is I suppose is OK for a database, but would at the very least would raise versioning issues for an ontology (less for OWL than OBO 1.2 I expect). Work arounds like using floats for rank orders seem rather non-ontological. Perhaps some sort of lattice that accommodates alternate orderings would suffice for this. I agree with Kevin and Jonathan that using rank assignments and ordering to validate taxonomic ordering is a valid use case, though I'm not sure how compelling I find it.
I think it would be useful to include provenance or authority information with rank terms since the number of such sources seems to still be growing.
Peter
Peter E. Midford
Mesquite Developer
Peter....@gmail.com
Jonathan Rees
> Richard,
> How do you deal with differences in usage between authorities (e.g., the respective bodies for Botany and Zoology)?
>
> Specifying a sort order field suggests that inserting intermediate level rank terms causes field values to change, which is I suppose is OK for a database, but would at the very least would raise versioning issues for an ontology (less for OWL than OBO 1.2 I expect). Work arounds like using floats for rank orders seem rather non-ontological. Perhaps some sort of lattice that accommodates alternate orderings would suffice for this. I agree with Kevin and Jonathan that using rank assignments and ordering to validate taxonomic ordering is a valid use case, though I'm not sure how compelling I find it.
Peter,
This is not at all what I was saying. I take as a given that you've
chosen the right design; I just don't understand what the design is.
If I did I would be able to answer the non-rhetorical questions about
correct use of the ontology that I posed.
If the answer is that there's no way to say anything incorrect using
these terms, no way that a database using these terms can contain a
mistake, then I'm not sure what it's good for. But I don't think this
is what you mean to imply, and I'm pretty sure that what you've done
is good for something.
As another example, suppose you published an article that said
"Peter's Taxon 12345 has rank Phylum", and then I published another
article that said "Peter's Taxon 12345 has rank Genus". And I cite
you and say that by "Peter's Taxon 12345" I mean whatever it is that
you mean by the phrase (same taxon concept). Are there any
circumstances under which my rank statement might be incorrect? (I
understand that we might be talking about a very nondiverse phylum,
but suppose we're not; or re-run this example where I say "Peter's
Taxon 12345 is not a Phylum".)
Jonathan
Richard Pyle
> How do you deal with differences in usage between
authorities (e.g.,
> the respective bodies for Botany and Zoology)?
I've dealt with it in two ways; neither of which I'm completely happy with.
One is to establish one set of Rank entities, and then define attributes of
ZoologicalLabel, BotanicalLabel, BacteriologicalLabel, etc. In that way the
same Rank entity is labeled as "Phylum" in the zoological context, and
"Division" in the botanical context (for example). The other alternative is
to do more or less what ITIS does, and define different sets of entities for
each Code-domain, so that there is a full set for Zoology, and another full
set for Botany (except, unlike ITIS, these are classed by Code, rather than
Kingdom).
I think I slightly prefer the first approach, and I can't remember what we
decided in the GNUB context (will revisit this soon).
Another problem is that the same term might be used in different senses, in
different disciplines (even within the same Code-domain). For example,
"Division" has one meaning in botany (that I am aware of), which is as
equivalent to the zoological Phylum. However, "Division" has been used in
at least two different ways within Zoology; either between Class and Order,
or as an infrageneric rank. This supports the idea of defining ranks
separately from the term used to represent the rank.
There is another issue involving ranks that is even more difficult to
resolve. That is the difference between ranks as abstract entities (i.e.,
"standard ranks"), and ranks as they have actually been used. The latter
needs to be dealt with when managing TaxonNameUsage TNU instances, because
it's often useful to capture the verbatim rank when one is given. There is
a wide diversity of things that have been branded as ranks over the past two
and a half centuries of taxonomic literature. However, for performing
simple reasoning (of the sort Kevin alluded to), it's usually much easier to
work with the abstract ("standard") ranks. I think for GNUB we decided that
each TNU would have both a TaxonRankID attribute, and a VerbatimTaxonRank
attribute; the former linking to a controlled vocabulary (actually, slightly
more than a controlled vocabulary, as described by Kevin), and the latter
would simply be a text string representing a UTF-8 representation of what
actually appeared on paper.
> Specifying a sort order field suggests that inserting intermediate level
rank
> terms causes field values to change, which is I suppose is OK for a
database,
> but would at the very least would raise versioning issues for an ontology
(less
> for OWL than OBO 1.2 I expect).
The way I've dealt with this is to leave gaps in the values assigned to the
Sort Order attribute, thereby allowing the insertion of newly encountered
ranks in-between established ranks, without changing the Sort Order value of
the established ones. This is a bit of a Kludge, of course, but I would
hope that, for standard ranks at least, we can establish a reasonably
complete set covering at least the abstract representation of every intended
rank, and then leave the variances of this to VerbatimRank.
> Work arounds like using floats for rank
> orders seem rather non-ontological.
I don't know ontology-space well enough to know if leaving gaps is similarly
non-ontological.
> Perhaps some sort of lattice that
> accommodates alternate orderings would suffice for this. I agree with
Kevin
> and Jonathan that using rank assignments and ordering to validate
taxonomic
> ordering is a valid use case, though I'm not sure how compelling I find
it.
I find it mostly useful for formatting compound name-strings according to
pre-set rules. These are obvious for ranks below Genus, but also apply to
higher ranks for things like qualifying names with "INCERAE SEDIS" and for
validating proper suffix formation on the name-string.
> I think it would be useful to include provenance or authority information
with
> rank terms since the number of such sources seems to still be growing.
In my mind, this is captured reasonably via VerbatimTaxonRank as an
attribute of TNU instances. I think the provenance for the "standard" ranks
ought to be an appropriate standards-setting body, such as TDWG, or the TCB
(http://www.bionomenclature.net/).
Aloha,
Rich
Hilmar Lapp
On Mar 31, 2011, at 5:01 PM, Jonathan Rees wrote:
> If the answer is that there's no way to say anything incorrect using
> these terms, no way that a database using these terms can contain a
> mistake, then I'm not sure what it's good for.
As Peter said earlier, it's an annotation property. It is as useful
(or not) as it is know for an ontology or an ontology class
(annotation) properties such as dc:creator, dc:contributor,
dc:lastModified, etc.
Annotation properties convey things about a class that in some
circumstances can be useful to know, but presumably not information
that is crucial to the class' semantics. In this sense, the idea that
the value of the has_rank property of a taxon class is part of its
semantics makes me shudder a bit - i.e., the notion that if I change
the has_rank property value with all else (type specimens, vouchers,
author, year, defining characters, etc) remaining unaltered I somehow
change the subject's semantics.
I understand the use-cases that have been stated here, but I think
using the taxonomic rank for any kind of inference over the semantics
of a taxon is using the wrong tool.
-hilmar
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- informatics.nescent.org :
===========================================================
Hilmar Lapp
On Mar 31, 2011, at 5:01 PM, Jonathan Rees wrote:
> If the answer is that there's no way to say anything incorrect using
> these terms, no way that a database using these terms can contain a
> mistake, then I'm not sure what it's good for.
As Peter said earlier, it's an annotation property. It is as useful
(or not) as it is know for an ontology or an ontology class
(annotation) properties such as dc:creator, dc:contributor,
dc:lastModified, etc.
Annotation properties convey things about a class that in some
circumstances can be useful to know, but presumably not information
that is crucial to the class' semantics. In this sense, the idea that
the value of the has_rank property of a taxon class is part of its
semantics makes me shudder a bit - i.e., the notion that if I change
the has_rank property value with all else (type specimens, vouchers,
author, year, defining characters, etc) remaining unaltered I somehow
change the subject's semantics.
I understand the use-cases that have been stated here, but I think
using the taxonomic rank for any kind of inference over the semantics
of a taxon is using the wrong tool.
-hilmar
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- informatics.nescent.org :
===========================================================
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- hlapp at drycafe dot net :
===========================================================
Erick Antezana
instead of as a 'class of taxa'?
what would be the pitfall(s) in introducing a 'subrank' relationship
among ranks? so that they get an ordering (maybe partial?)
genus is_subrank_of family
species is_subrank_of genus
Erick
Hilmar Lapp
On Apr 1, 2011, at 4:35 AM, Erick Antezana wrote:
> so it means that a rank is better conceptualised as a "metaclass"
> instead of as a 'class of taxa'?
In my understanding a meta-class is a class that is also an
individual. If we use the rank terms directly for OWL annotation
property values, then I guess indeed that would be a proper way to
call them.
> what would be the pitfall(s) in introducing a 'subrank' relationship
> among ranks? so that they get an ordering (maybe partial?)
>
> genus is_subrank_of family
> species is_subrank_of genus
Sure, you could do that (and presumably make is_subrank_of
transitive?). The main danger (in my view anyway) is in using the
has_rank property of a taxon class to infer something about the
semantics of the set of the taxon class' members.
-hilmar
--
===========================================================
Peter Midford
Unfortunately, apart from the simple punning facility that OWL2 added, metaclasses will take you out of the realm of DL-logics, which means reasoning might become intractable.
Agree with Hilmar that you could do this. I would recommend doing that locally - take the terms from a rank vocabulary and add the subrank_of relations as local (to your ontology or project). Many of the messiest problems come from making statements relating ranks from different authorities, but if all your ranks are botanical, you could axiomatize the botanical hierarchy and even make it available as a resource for botantical ontologies. However, I guess I'm still waiting for the compelling use case - and I agree there is a danger in inferring anything non-trivial about taxa on the basis of their rank.
Peter
Peter E. Midford
Mesquite Developer
Peter....@gmail.com
------------------------------------------------------------------------------
Pankaj Jaiswal (OSU)
it nothing but a simple flat list of taxon names as subsets/children.
This may appeal to CS and Ontology world but may not attract biologists
unless we are looking ahead for a way to say that taxonomy in the coming
5-10 years will become more flexible/plastic. Because as we gather more
evidences and sequence based alignments matched with morphology and
growth habit, taxonomy will be redefined from its current Linnaean
classification format. On the other hand is there a reason we cant use
ranks as relationships.
e.g.
taxon-A is_a_species_of taxon_B is_a_genus_of taxon_B
Pankaj
Stefan Schulz
Schulz S, Stenzhorn H, Boeker M.
The ontology of biological taxa.
Bioinformatics. 2008 Jul 1;24(13):i313-21.
PMID: 18586729 [PubMed - indexed for MEDLINE] Free PMC Article
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2718636/?tool=pubmed
Best regards,
Stefan Schulz
2011/4/1 Pankaj Jaiswal (OSU) <jais...@science.oregonstate.edu>:
--
Stefan SCHULZ (Univ.-Prof. Dr. med.)
Institut für Medizinische Informatik,
Statistik und Dokumentation
Medizinische Universität Graz
Auenbruggerplatz 2/V
8036 Graz (Austria)
http://www.medunigraz.at/imi
+43 (0)316 385 16939
+43 (0)316 385 13201
http://purl.org/steschu
mailto:stefan...@medunigraz.at
Skype: stschulz
[ home: Eschholzstrasse 70
[ 79115 Freiburg (Germany)
[ mobile: +49(0)176 23938735
Jonathan Rees
>
> On Mar 31, 2011, at 5:01 PM, Jonathan Rees wrote:
>
>> If the answer is that there's no way to say anything incorrect using
>> these terms, no way that a database using these terms can contain a
>> mistake, then I'm not sure what it's good for.
>
> As Peter said earlier, it's an annotation property. It is as useful (or not)
> as it is know for an ontology or an ontology class (annotation) properties
> such as dc:creator, dc:contributor, dc:lastModified, etc.
Dublin Core properties have a *lot* of meaning - so much that I could
in principle be charged with fraud by using DC statements to claim as
my own a paper that wasn't mine. So annotation != meaningless. My
question was just what meaning, or particular use, the rank terms had.
I *know* they have meaning, because I know Peter wouldn't waste his
time on something meaningless. I just don't know what it is, and was
hoping to draw him out so that it could be documented on the wiki.
> Annotation properties convey things about a class that in some circumstances
> can be useful to know, but presumably not information that is crucial to the
> class' semantics. In this sense, the idea that the value of the has_rank
> property of a taxon class is part of its semantics makes me shudder a bit -
> i.e., the notion that if I change the has_rank property value with all else
> (type specimens, vouchers, author, year, defining characters, etc) remaining
> unaltered I somehow change the subject's semantics.
>
> I understand the use-cases that have been stated here, but I think using the
> taxonomic rank for any kind of inference over the semantics of a taxon is
> using the wrong tool.
That is not at all what I was suggesting. I was just talking about how
a developer or author would use this property, in practice. My
question had nothing to do with automated inference or OWL or anything
like that. To try to draw Peter out, I asked some very simple and
concrete questions - which he didn't answer, and I'm not sure why.
If I can't get my meaning across to Peter and you in email - and I
know and respect you both, so I know you would be able to understand
something that was stated clearly - then I have failed in a way that
baffles me, and I will go off for a while and try to figure out why I
am being confusing. I think I will give up for now, and wait until I
next see Peter in person to continue. Email is probably not conducive
to this particular discussion.
Best
Jonathan
Hilmar Lapp
On Apr 2, 2011, at 7:43 PM, Jonathan Rees wrote:
>> I understand the use-cases that have been stated here, but I think
>> using the
>> taxonomic rank for any kind of inference over the semantics of a
>> taxon is
>> using the wrong tool.
>
> That is not at all what I was suggesting.
I know. But rank-based taxonomy is a minefield :-)
> I was just talking about how a developer or author would use this
> property, in practice.
And that's a good and valid question. I can tell you how we use it
within Phenoscape: solely for display purposes on the web-interface
(kb.phenoscape.org). I.e., it is useful to know what rank a taxon with
phenotype annotations has. That's why I call it an annotation
property. I agree with you that annotation property doesn't mean it's
a free for all.
Does this help at all in answering your question?
-hilmar
--
===========================================================
: Hilmar Lapp -:- Durham, NC -:- hlapp at drycafe dot net :
Peter Midford
It wasn't my intention to leave you hanging. I'm not sure the answer I can give will satisfy you, but I'll give it a try. TAXRANK primarily came about because TTO had originally followed the approach Chris Mungall took with the NCBI taxonomy ontology which had (and NCBI continues to have) two trees in the ontology, one for taxa, one for ranks. Although that isn't best practice, if NCBI is the only taxonomy you use, it doesn't seem too harmful. However, Phenoscape has plans to expand to vertebrates, but have separate taxonomies for the major vertebrate groups (revamping the ATO is part of this effort). For supporting our multiple, non-overlapping taxonomies, keeping ranks in a separate resource is clearly the way to go, which is why TAXRANK exists.
The reason we haven't added more semantics to rank terms is both because we haven't had a phenoscape use case that requires it (we really only use them for display purposes at present), but also because providing a set of names allows those who need the additional semantics to build ontologies that reflect the taxonomic thinking of their particular domain authority (zoology, botany, bacteriology, etc.). There's enough incongruence among the authorities that trying to capture it all in a single ontology would, I expect, require at least a workshop to capture all the issues, apart from the engineering effort to make a consistent ontology that behaved properly across authorities. It's even messier than that, as there is a term in TAXRANK that seems specific to Drosophilists. But I'm sure this incongruence isn't news to you. Since we didn't have a strong use case, I didn't give much consideration to half-way measures such as asserting ordering relations among the standard Linnaean ranks.
Note that has_rank is defined (as an annotation property / OBO metadata tag) in TTO, not TAXRANK. This allows the property linking taxa and ranks to be defined as needed in a particular taxonomic ontology. I expect Phenoscape will be experimenting again with individual-based taxonomies in near future as part of transitioning to OWL and allow a reconsideration of our linking property.
I think TAXRANK can provide a common baseline vocabulary for projects whether or not they see the need to define rank ordering relations. If enough projects decide to build ontologies that order ranks and there is a desire to attempt to integrate across authorities, I expect the Phenotype ontology RCN to welcome proposals for a integration workshop when the time comes.
I suspect this is less explanation that you were hoping for, but this is where they stand. (I see Hilmar has already responded).
Peter
Peter E. Midford
Phenoscape Ontology Curator
Peter....@gmail.com
------------------------------------------------------------------------------