HCLSIG BioRDF Subgroup/Meetings/2009-06-22 Conference Call
From W3C Wiki
- Date of Call: Monday June 22, 2009
- Time of Call: 11:00 am Eastern Time
- Dial-In #: +1.617.761.6200 (Cambridge, MA)
- Dial-In #: +33.4.89.06.34.99 (Nice, France)
- Dial-In #: +44.117.370.6152 (Bristol, UK)
- Participant Access Code: 4257 ("HCLS")
- IRC Channel: irc.w3.org port 6665 channel #hcls (see W3C IRC page for details, or see Web IRC)
- Duration: ~1 hour
- Frequency: bi-weekly
- Convener: Kei Cheung
- Scribe: M. Scott Marshall
Kei Cheung, M. Scott Marshall, Lena Deus, Jun Zhao, Matthias Samwald
Adrian Paschke, Rob Frost
- Introduction and Roll Call (Kei)
- TCM data -- DILS 09 poster/demo (Jun)
- Query federation extension/expansion (All)
- aTags (Matthias)
<kei> Kei: went over agenda items
<mscottm> Kei: recently had a Skype about TCM with Jun, Matthias, and Anja
<mscottm> .. about TCM, Jun can provide an update
<mscottm> ..Submitted abstract to DILS2009
<mscottm> Jun: Have conferred with others on the poster
<mscottm> ..checked genetic connections between TCM and LODD
<matthias_samwald> aTags about TCM: http://hcls.deri.org/atag/data/tcm_atags.html
<mscottm> ..will have to move to a faster server after testing during Skype session
<junzhao> Silk: http://www4.wiwiss.fu-berlin.de/bizer/silk/
<mscottm> ..consulted with Anja who uses Silk to connect TCM and LODD
<mscottm> ..looked at how many direct links can be used, need to form a strategy of when to use direct links,
<mscottm> ..want to study the overlap of the mappings to provide confidence in the mappings
<mscottm> Kei: There are High Throughput network data sets such as the Diseasome and TCM that might be interesting to explore to find where they have common disease genes.
<mscottm> ..That's one possible angle. On the other hand, the data sets represent different views from Traditional Chinese Medicine and Western drugs.
<mscottm> ..the data sets could possibly provide complementary information, for example, how traditional herbs could be used to help treat chemotherapy side effects.
<mscottm> Lena: the mapping is done mostly manual right?
<mscottm> Kei: Yes, mostly manual, but it could provide insight into automation.
<mscottm> Scott: I saw Huajun Chen's presentation in Amsterdam. It seems necessary to map (Western) disease names to the sets of symptoms that are necessary to diagnose conditions in Chinese Medicine.
<junzhao> do we have HuaJun's presentation online somewhere?
<Kei> It may be worth inviting Huajun to give a talk/discussion here.
<mscottm> Kei: Yes, such a mapping would be useful.
<mscottm> Jun: That would be interesting. In the current TCM dataset, we don't have the symptoms as they are named in traditional chinese medicine.
<mscottm> Kei: It could be interesting to connect to Huajun's databases to establish a a more comprehensive resource.
<junzhao> Lena: are the mappings available through SPARQL endpoints yet?
<junzhao> Jun: we are hosting them through SPARQL endpoints and will release them to public once they are finalized and send an email to the mailing list
<mscottm> Jun: We haven't released the data sets yet.
<mscottm> Kei: There are some Chinese labels in the TCMGeneDIT that could be used.
<mscottm> Scott: Might it be interesting to provide a way to view either Chinese or English labels of some data?
<mscottm> Jun: The Chinese version of DBpedia has more information in the TCM area than the English version so it could be interesting to host the Chinese DBpedia. However, this would be expensive so a pilot would be the best approach.
<mscottm> Kei: How to continue forward with federation? Possibly look at rules and access control.
<mscottm> Kei: Scott, do you know about the status of BMC special issue?
<mscottm> Scott: Status is ongoing but I don't know precisely when it will come out.
<LenaDeus> Scott: using rules to rewrite the SPARQL queries
<LenaDeus> Kei: Rules can have different granularity, it would be good to be able to use metadata to trim the results
<mscottm> tx Lena!
<mscottm> Kei: example: use rules to decide which part of the brain expresses certain genes
<LenaDeus> Scott: Regarding worrkflows, in mark wilkinson's work, the ediscrete and continuous domain are integrated by description logics
<mscottm> Lena: Mark W. translates predicate in SPARQL query to a call to a service
<LenaDeus> Kei: what other type of rule examples can be integrated in the query feredation?
<LenaDeus> Kei: Some concepts applied in Mark's tool can be applied here as well.
<mscottm> Kei: microarray data, rich in annotation, could benefit from rules as well
<mscottm> Lena: making all microarray data in triples is expensive
<mscottm> Kei: You could use the list of differentially expressed genes.
<mscottm> ..we can identify projects, and access papers via PMID's to extract lists of genes.
<mscottm> Kei: If there is interest in the group, I would be willing to investigate the resources available for neuroscience gene expression data.
<LenaDeus> Mathia: curated literature that describes actions used in traditional medicine and annotated with aTags;
<mscottm> Matthias: for aTags the focus is twofold: create aTags from literature and capturing biomedical knowledge for efficient query.
<LenaDeus> Kei: is it still a manual task to do the aTag curation?
<matthias_samwald> NCBO Biomedical Annotator, Terminizer
<mscottm> Matthias: Looking for a tag recommender technique, currently at OBO annotator.
<LenaDeus> (i have to go, thank you all :) )
<kei> scott: expressed interest in connecting aida with atags
<kei> scott: coming up with atag recommendation is of general interest to me
<kei> scott: term recommendation including ranking and storing atags in triplestores
<kei> scott: making effective term recommendation is particularly interesting
<mscottm> Matthias: I have extended that Science Commons text annotator with a few new features
<mscottm> .. can go to the service with a PMID and it will return the article annotated with aTags.
<mscottm> ..also setup a wiki about this
<matthias_samwald> http://terminizer.org/ is also made in manchester
<mscottm> Scott: NACTEM has a data source of PAS predicate argument structures - much like triples in a sense.