[Obo-format] xrefs to MeSH
David Osumi-Sutherland
We're in the process of mapping FlyBase publication type terms to MeSH. This makes us consistent with PubMed - the main source of our biblio data.
Ideally I'd like to be able to make EquivalentClass links, but presumably I'd need an up-to-date version of MeSH in OBO or OWL for that to be viable ? If so, does anyone know if there is one available?
In the absence of this, I've been looking at using the OBO xref mechanism (which is also useful for attributing definitions.)
Unfortunately, xrefs to MeSH are somewhat problematic. Here's the GO.xrf_abbs entry:
abbreviation: MeSH
database: Medical Subject Headings
object: MeSH heading
example_id: MeSH:mitosis
generic_url: http://www.nlm.nih.gov/mesh/2005/MBrowser.html
url_syntax: http://www.nlm.nih.gov/cgi/mesh/2005/MB_cgi?mode=&term=[example_id]
url_example: http://www.nlm.nih.gov/cgi/mesh/2005/MB_cgi?mode=&term=mitosis
Note the year embedded in the URL. It seems that the name/year combination gives a stable reference. But to use this, we would need a new MeSH xref abbreviation for each year. There's also a further wrinkle to this:
"Cell Transformation, Neoplastic"
>
http://www.nlm.nih.gov/cgi/mesh/2010/MB_cgi?mode=&term=Cell+Transformation,+Neoplastic
This feels unsatisfactory, but I'm not sure there is any other good option.
MeSH terms have unique IDs, but as far as I can gather, there is no way to use these to roll a URL (although you can query them from here http://www.nlm.nih.gov/mesh/2010/mesh_browser/MBrowser.html).
Tree node IDs have been used by GO curators, e.g.:
id: GO:0006927
name: transformed cell apoptosis
namespace: biological_process
def: "The process of apoptosis in transformed cells, cells that have undergone changes manifested by escape from control mechanisms, increased growth potential, alterations in the cell surface, karyotypic abnormalities, morphological and biochemical deviations from the norm." [GOC:jl, MeSH:C04.697.152]
These can be used to roll URLs in combination with a year and term name:
i.e.- they appear to only be useful for specifying which, of multiple possible tree nodes for a term to display on linking.
So - can we have year specific xrefs for MeSH in order to get reliable expansions? It has the advantage of recording version information for free. Or does anyone have a better suggestion?
Cheers,
David
David Osumi-Sutherland, PhD
Ontologist
FlyBase / Virtual Fly Brain
Department of Genetics,
University of Cambridge,
Downing Street,
Cambridge, CB2 3EH, UK
Tel: +44 (0)1223 333 963
Fax: +44 (0)1223 766 732
http://www.virtualflybrain.org
------------------------------------------------------------------------------
Cloud Computing - Latest Buzzword or a Glimpse of the Future?
This paper surveys cloud computing today: What are the benefits?
Why are businesses embracing it? What are its payoffs and pitfalls?
http://www.accelacomm.com/jaw/sdnl/114/51425149/
_______________________________________________
Obo-format mailing list
Obo-f...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-format
Melissa Haendel
eagle-i Networking Research Resources
Oregon Health & Science University
hae...@ohsu.edu
skype: melissa.haendel
David Osumi-Sutherland
Obo-discuss mailing list
Obo-d...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-discuss
Alan Ruttenberg
> Hi David,
> not sure if this is what you are looking for, but we've been able to get a
> MeSH taxonomy out of NCBO for use as a reference taxonomy in eagle-i using
> their view extractor tool:
> http://www.bioontology.org/wiki/index.php/View_Extraction
>
> The classes look like this:
> <!-- http://purl.bioontology.org/ontology/MSH/D000008 -->
>
> <owl:Class rdf:about="http://purl.bioontology.org/ontology/MSH/D000008">
> <rdfs:label>Abdominal Neoplasms</rdfs:label>
> <rdfs:subClassOf
> rdf:resource="http://purl.bioontology.org/ontology/MSH/D009371"/>
> </owl:Class>
The MeSH terms are not related by subClassOf, they constitute a
terminology and should be related by the skos broader* etc
relationships.
Moreover the MeSH version on bioportal seems to be an extraction from
UMLS rather that primary source. As such the relation names as
displayed are ... obscure.
Only a careful review would let one assess whether some of the
relations warrant being subclass relations.
If David is only using the publication terms then it shouldn't matter to him.
We have a neurocommons MeSH version that is based on an earlier skos
version, but also creates terms for each term/qualifier pair (since
these tend to have different meanings. I would recommend that or
another version that builds upon it be used. See
http://neurocommons.org/page/Bundles/mesh/mesh-skos
A lasting rendering of MeSH suitable for use within the OBO Foundry,
would, I think, be one based on IAO, as MeSH is an indexing resource,
rather than a referring ontology. It would take care with qualified
terms, and have some theory of how to manage reference to terms in
specific yearly versions. My recommendation for simple xref,
would be (in this order)
- A version of MeSH based on IAO taking into account the issues noted
- A version of MeSH based on the Neurocommons version, which is
SKOS-based, takes care with qualifiers, and is built direct from MeSH
source, but not the versioning issues.
- A direct link to the MeSH web space.
- The current bioportal version (this falls under the line of recommended)
-Alan
David Osumi-Sutherland
OK. Given that I need something now, this looks like a good way to go for the semantics. I'm clear on how to do this in OWL
In Manchester syntax: <http://purl.obolibrary.org/obo/FBcv_0000207> equivalentTo <http://purl.org/commons/record/mesh/D016422>
And we can now express equivalence in OBO.
However, it's not clear to me how we deal with the full URL for the MeSH term in OBO though - a question for Chris Mungall I think.
That deals with the semantics, but given that these commons purls don't resolve to anything useful, I think it's also worth having an xref in place that expands to something that does. It's probably sufficient for this to be linked to the definition. This could be the bioportal purl, or a direct link to the MeSH site. I'd prefer the latter, but it seems like only the way we get this is with year specific GO.xrf_abbs entry.
Cheers,
David
> Obo-discuss mailing list
> Obo-d...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obo-discuss
David Osumi-Sutherland, PhD