Alan Ruttenberg wrote:
On Feb 12, 2008, at 3:58 PM, Kei Cheung wrote:
Hi Olivier, Alan, et al,
A prototype version of our neuroscience semantic web portal is
accessible at: http://neuroweb3.med.yale.edu
Nice :)
Thanks :-) . There is still a lot of room for improvement.
Bear in mind, this is a working prototype for proof of concept and
it is still under active development. The Web interface has 3
panels: search panel (on top), facet panel (below the search panel
on the left) and data panel (below the search panel on the right).
The user can enter a key word and select a facet and data sources
(ontologies) against which the key word is searched. The search
results are displayed and organized based on the selected facet.
Each leaf node (neuron, neuron property, gene, or drug target
(receptor)) of the facet tree can be selected for retrieving more
detailed information from each of the selected data sources.
Currently, we have implemented 4 facets: neuron (default), neuron
properties, gene, and drug. While the first two facets were created
based on the SenseLab and CCDB ontologies, gene facet was created
based on GO's molecular functions and the drug facet was created
based on ChEBI.
I think MeSH can potentially be a source for creating new facets
(e.g., diseases, drug actions, etc) ...
If you can supply a list of genes, you can retrieve a set of mesh
terms by linking via entrez gene to pubmed, then to mesh. Same if you
supply a list of pubmed ids. You can also limit to a specific branch
of mesh easily enough. If that's the sort of think you had in mind,
let me know and I can construct the query, or Matthias can.
Thanks for bringing this up. I have also thought about that. The only
thing is that I'd like to spend a little bit time to think about how to
constraint the MeSH facet tree in a meaningful context.
Also, I have a question regarding retrieval of entez gene
information including gene ids, symbols, and snyonyms. Is such
information available from the HCLS KB and/or RDF entrez gene
dataset that Olivier's group had created?
Separate email.
For educational purposes, perhaps have an option to show the sparql
queries you use to retrieve the information and offer a sparql
endpoint to let people play with fiddling the queries.
Although we have everything loaded into Oracle triplestore at the
moment, we have also thought about the possibility of using sparql
endpoints to query other triplestores live.
I think I just got a flu, need to go offline ...
-Kei
Regards,
Alan
Cheers,
-Kei
Alan Ruttenberg wrote:
On Feb 11, 2008, at 3:34 PM, Olivier Bodenreider wrote:
eric neumann wrote:
I haven't been good at following the calls lately, but still am
sympathetic to the endeavor and committed to helping as I can
(just not quite efficiently).
Starting with MeSH might be a good idea as it it does not have the
intellectual restrictions inherent to other sources in the UMLS.
Moreover, MeSH is available in XML and could be easily transformed
into RDF, allowing for a possible resolution of the URIs.
There is already one RDF translation in the KB (script from Van
Assem et. al., modified by Jonathan). This is in to SKOS. There's
also a prototype uri scheme that we used in the demo, if you are
looking for a model.
-Alan
-- Olivier
Eric
On Feb 11, 2008 9:58 AM, Matthias Samwald <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
I have also added several ideas to the wiki page [1]:
* OBO Structured Digital Abstracts
* SIOC for Science
* OntoWiki and Semantic MediaWiki
* HCLS KB Decentralisation
* Define a simplified OWL-to-RDF mapping for a subset of OWL
Cheers,
Matthias Samwald
Semantic Web Company / DERI Galway / Yale Center for Medical
Informatics
[1] http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup/ Brainstorming
