Adding addition OWL & RDF artifacts to the NC RDF repository

On May 24, 2007, at 11:36 AM, William Bug wrote:

> Hi Alan,
>
> Here's the GO/NCBO/Berkeley list that Chris Mungall
curates (a page  
> I'm certain you've been to many times):
> 	http://www.ber
keleybop.org/ontologies/
>
> As you know, it contains - among a host of other
XML-based formats  
> - XML-serialized OWL & RDF versions of most all OBO
ontologies, as  
> well as an extensive collection of mappings (also
available in RDF  
> & OWL).
>
> Rather than going through and copy & pasting ALL
the OWL & RDF  
> links into a text file, I thought I just send this
reminder and cc  
> to Chris, as he may have other ideas about how to get
these into  
> the NC repository that don't include error-prone copy
& pasting or  
> require someone to write a scraper for this page.

NC? My acronym buffer is full - do I know this one?

> It would also be worth other members of HCLS IG
pitching in to help  
> give you some sense of priorities for these files.  I
could give  
> you my sense of that - but that may be less inclusive
than we  
> really want for future HCLS IG demos or NC efforts in
general.
>
> In terms of scraping the page, everything is pretty
regularized  
> (and probably algorithmically generated, if I know
Chris).
>
> For the ontologies themselves, the URL template appears
to be:
> 	http://www.berkeleybop.org/ontologies/obo-al
l/<ontology_name>/ 
> <ontology_name>.owl
> 	http://www.berkeleybop.org/ontologies/obo-al
l/<ontology_name>/ 
> <ontology_name>.rdf

The RDF is pretty much legacy pseudo-RDF, from our first
attempt to  
do GO in RDF many years ago - I would ignore this

You can also fetch by ID-space prefix; for example

<BASE-URL> /owl/ <IDspace>

There isn't always a 1-1 mapping; for example GO is one
IDspace and 3  
ontologies. So

<BASE-URL>
/obo-all/biological_process/biological_process.owl

is a subset of

<BASE-URL> /owl/GO

<BASE-URL> can be either purl.org/obo or http://www.berkeleybop.or
g/ 
ontologies. The former is the basis for the stable URIs as
it is  
independent of hosting location

> For the inter-ontology maps, the URL template appears
to be:
> 	http:/
/www.berkeleybop.org/ontologies/obo-all/ 
> <ontology1_xp_ontology2>/<
ontology1_xp_ontology2 >.owl
> 	ht
tp://www.berkeleybop.org/ontologies/obo-all/<  
> ontology1_xp_ontology2 >/< ontology1_xp_ontology2
>.rdf
>
> 'xp' for 'cross-product', I assume.

yes, 'cross product' is the name that seems to have stuck
for this  
kind of intra and inter ontology combinatorial definitions.

it's not quite as regular as you make out - for example, it
is just  
mammalian_phenotype_xp because  
mammalian_phenotype_xp_pato_xp_go_xp_chebi_xp_mouse_anatomy
would be  
too much

Note for the xps, there is also a separate imports file;

<BASE-URL> / <a_xp_b> /
<a_xp_b>.imports.{owl,obo}

eg

http://www.berkeleybop.org/ontologies/obo-all/go_xp_ch
ebi/ 
go_xp_chebi.imports.owl

Which is nothing but a collection of import statements. I
realise  
that this is a different strategy from the semweb norm
whereby you  
include the import in the main owl file, but I think in
general its  
easier to do late coupling than late decoupling hence the
decision.  
Also I'm not keen on mixing imperative parser directives in
with  
declarative ontological statements.

be warned these xp files are at the draft stage, reasoning
over them  
will give you some unusual results

> So to create a priority list, we'd just need the name
of the  
> ontology.  Everything else could be done
algorithmically.
>
> Cheers,
> Bill
>
> Bill Bug
> Senior Research Analyst/Ontological Engineer
>
> Laboratory for Bioimaging  & Anatomical
Informatics
> www.neuroterrain.org
> Department of Neurobiology & Anatomy
> Drexel University College of Medicine
> 2900 Queen Lane
> Philadelphia, PA    19129
> 215 991 8430 (ph)
> 610 457 0443 (mobile)
> 215 843 9367 (fax)
>
>
> Please Note: I now have a new email - William.BugDrexelMed.edu
>
>
>
>

> Rather than going through and copy & pasting ALL
the OWL & RDF links  
> into a text file, I thought I just send this reminder
and cc to  
> Chris, as he may have other ideas about how to get
these into the NC  
> repository that don't include error-prone copy &
pasting or require  
> someone to write a scraper for this page.

I think that we probably do not want to use (and even
import) all of the ontologies in the current OBO collection.
We should probably focus on the OBO Foundry ontologies and
some of the more consistent OBO Foundry candidates. Some of
the ontologies would need some remodelling before they can
be integrated into a consistent demo scenario in a really
useful way.
For example, the evidence code ontology could better be
represented as an ontology of processes of biomedical
experiments and techniques.

-- Matthias
-- 
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu
sparen! 
Ideal für Modem und ISDN: http://www.gmx.n
et/de/go/smartsurfer

> [2] For instance, definitions in the PATO OWL file
(quality.owl) are  
> not visible in Protege.  The OWL version of PATO has an
 
> AnnotationProperty called
"oboInOwl:hasDefintion" that has the value  
> "_:A4843".  

I and some other people have already nagged Chris about the
representation of defintions and synonyms in the OWL
versions of OBO ontologies.
This is partly related to a larger problem, namely the
inhomogeneous representation of labels, descriptions and
definitions among biomedical Semantic Web ontologies. I have
started a Wiki page on this topic:

h
ttp://esw.w3.org/topic/HCLS/Labels_and_Definitions

The page is intended to contain the following (please add to
it / discuss!):

* Examples from biomedical ontologies of non-standard
constructs to represent Labels, names, descriptions and
definitions of entities.

* An analysis of the motivations behind the creation of
these constructs

*A description of the problems that arise through a lack of
harmonization (e.g. for queries and user interfaces)

* A review of the basic constructs in the RDF, RDFS and OWL
vocabularies and their intended usage for labels,
descriptions and definitions

* An informal recommendation for the representation of
labels, descriptions and definitions and suggestions for the
harmonization of biomedical ontologies in this regard.


I have started comparing the constructs from a few
ontologies, but there are many additional ontologies that
have created their own style for the representation of this
information. We should try to find the smallest common
denominator between all these approaches, because this
really is a large hindrance to practical interoperability
between our ontologies.

cheers,
Matthias Samwald

----------

Yale Center for Medical Informatics, New Haven /
Section on Medical Expert and Knowledge-Based Systems,
Vienna /
http://neuroscientific.net




.
-- 
GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free
SMS.
Alle Infos und kostenlose Anmeldung: http://www.gmx.net/
de/go/freemail

On May 25, 2007, at 8:44 AM, samwaldgmx.at wrote:

>> [2] For instance, definitions in the PATO OWL file
(quality.owl) are
>> not visible in Protege.  The OWL version of PATO
has an
>> AnnotationProperty called
"oboInOwl:hasDefintion" that has the value
>> "_:A4843".

note there is a protege3 plugin that has oboinowl aware
displays

..but of course this doesn't help if you're using  p4,
swoop, lsw,  
triple20 on any other sw tool

> I and some other people have already nagged Chris about
the  
> representation of defintions and synonyms in the OWL
versions of  
> OBO ontologies.

nagging me directly won't do much, it's a collaborative
effort. the  
best place to comment is probably here:
obo-formatlists.sourceforge.net

my own preferred way of doing this is to use simple
class-label  
annotprops and reify them to attach provenance info. this
would  
probably be even less popular (even *less* support in owl
tools), but  
it makes sense to me. See also the reif/NG discussion on
this list -  
NGs would seem clunky here

> This is partly related to a larger problem, namely the 

> inhomogeneous representation of labels, descriptions
and  
> definitions among biomedical Semantic Web ontologies.

++ to that

> I have started a Wiki page on this topic:
>
> h
ttp://esw.w3.org/topic/HCLS/Labels_and_Definitions
>
> The page is intended to contain the following (please
add to it /  
> discuss!):
>
> * Examples from biomedical ontologies of non-standard
constructs to  
> represent Labels, names, descriptions and definitions
of entities.
>
> * An analysis of the motivations behind the creation of
these  
> constructs
>
> *A description of the problems that arise through a
lack of  
> harmonization (e.g. for queries and user interfaces)
>
> * A review of the basic constructs in the RDF, RDFS and
OWL  
> vocabularies and their intended usage for labels,
descriptions and  
> definitions
>
> * An informal recommendation for the representation of
labels,  
> descriptions and definitions and suggestions for the
harmonization  
> of biomedical ontologies in this regard.
>
>
> I have started comparing the constructs from a few
ontologies, but  
> there are many additional ontologies that have created
their own  
> style for the representation of this information. We
should try to  
> find the smallest common denominator between all these
approaches,  
> because this really is a large hindrance to practical 

> interoperability between our ontologies.

seems worthwhile. but if you want it to take off why
restrict it to  
the HCLS?

> cheers,
> Matthias Samwald
>
> ----------
>
> Yale Center for Medical Informatics, New Haven /
> Section on Medical Expert and Knowledge-Based Systems,
Vienna /
> http://neuroscientific.net

>
>
>
> .
> -- 
> GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free
SMS.
> Alle Infos und kostenlose Anmeldung: http://www.gmx.net/
de/go/freemail
>

On May 25, 2007, at 8:44 AM, samwaldgmx.at wrote:

>
>> [2] For instance, definitions in the PATO OWL file
(quality.owl) are
>> not visible in Protege.  The OWL version of PATO
has an
>> AnnotationProperty called
"oboInOwl:hasDefintion" that has the value
>> "_:A4843".
>
> I and some other people have already nagged Chris about
the  
> representation of defintions and synonyms in the OWL
versions of  
> OBO ontologies.
> This is partly related to a larger problem, namely the 

> inhomogeneous representation of labels, descriptions
and  
> definitions among biomedical Semantic Web ontologies. I
have  
> started a Wiki page on this topic:
>
> h
ttp://esw.w3.org/topic/HCLS/Labels_and_Definitions
>
> The page is intended to contain the following (please
add to it /  
> discuss!):
>
> * Examples from biomedical ontologies of non-standard
constructs to  
> represent Labels, names, descriptions and definitions
of entities.
>
> * An analysis of the motivations behind the creation of
these  
> constructs
>
> *A description of the problems that arise through a
lack of  
> harmonization (e.g. for queries and user interfaces)
>
> * A review of the basic constructs in the RDF, RDFS and
OWL  
> vocabularies and their intended usage for labels,
descriptions and  
> definitions
>
> * An informal recommendation for the representation of
labels,  
> descriptions and definitions and suggestions for the
harmonization  
> of biomedical ontologies in this regard.

I have some *experimental* code that does a reasonable job
of  
harmonizing the following 3 styles of annotationprops: bfo,
birn and  
nci. see

http://www.be
rkeleybop.org/obo-conv.cgi

it is primarily for owl<->obo -
to harmonize from owl to owl you'd have to do a double
transform  -  
and it would all be harmonized to the oboinowl style - but
this could  
potentially be modified if there was interest

> I have started comparing the constructs from a few
ontologies, but  
> there are many additional ontologies that have created
their own  
> style for the representation of this information. We
should try to  
> find the smallest common denominator between all these
approaches,  
> because this really is a large hindrance to practical 

> interoperability between our ontologies.
>
> cheers,
> Matthias Samwald
>
> ----------
>
> Yale Center for Medical Informatics, New Haven /
> Section on Medical Expert and Knowledge-Based Systems,
Vienna /
> http://neuroscientific.net

>
>
>
> .
> -- 
> GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free
SMS.
> Alle Infos und kostenlose Anmeldung: http://www.gmx.net/
de/go/freemail
>
>