HCLSIG/PharmaOntology/Meetings/2010-09-09 Conference Call
Conference Details
* Date of Call: Thursday September 9 2010 * Time of Call: 12:00pm - 1:00pm ET * Dial-In #: +1.617.761.6200 (Cambridge, MA) * Dial-In #: +33.4.26.46.79.03 (Paris, France) * Dial-In #: +44.203.318.0479 (London, UK) * Participant Access Code: 42572 ("HCLS2") * IRC Channel: irc.w3.org port 6665 channel #HCLS2 (see W3C IRC page for details, or see Web IRC) * Mibbit quick start: Click on mibbit for instant IRC access * Duration: 1h * Convener: Michel
Agenda
- Next Steps/Strategy Discussion/doodle poll - Michel
- Completing the Demo - Michel
- AOB
Minutes
- thanks to EricP and BobP for scribing
- attending: michel, susie, elgar, eric p, bob, matthias, christi, scott
-> use cases http://doodle.com/r?url=http%3A%2F%2Fesw.w3.org%2FHCLSIG%2FPharmaOntology%2FUseCases -> TMO directions Doodle Poll http://doodle.com/28fvi74h6atyktyd
michel: some are fleshed out; some not
... created a doodle poll to ask folks where they want to push
Elgar: There is one case, integrated informatics use case
... do people want to consider that separately?
Michel: That seems to be more like clinical use case.
michel: that looks like "clinical research category" : patient data, trials, molecular details
Susie: hesitating 'cause these are broad
Susie: These are broad headings, depends on what is done w/in each part
... integrating should be considered b/c that's what we do
Elgar: Pharmacogenomics, omics, haven't made up my mind yet
Michel: discussing what's under each of the omics
ericP notes that http://doodle.com/28fvi74h6atyktyd remains nicely iconic
... look at genetic variation, how it affects response. (not quit getting all this!)
... talking w. Anya, extremely busy w next LODD map
... will discuss plan for regularity of update of datasets
Michel: Can also discuss patient data
Q: Are the LODD data important for us?
Michel: Will make scripts available for conversion of datasets
So, for the last papers we relied on LODD + Michel's data; will we keep that model, w Anya's data?
Michel: Anya will use Silk
... want LODD to produce separate file sets
... so it can become part of the workflow
... Question is how accurate in LODD, since using lexical mapping
... Susie, what about quality, cross-linking
Susie: We do have some questions about linking quality
... we need to undertake some work for LODD link quality
Michel: Don't think that some of the LODD links may be inconsistent, we need to look at when we update the paper
Michel: Part of the value of TMO could be checking data consistency as the maps go forward
Matthias: Need to check how complicated this would be, but yes will look at.
Action Item: matthias to look into OWLIM to reason about TMO + mappings + data
Action Item: michel to discuss with Anja the update workflow for LODD datasets, and inquire about SILK mapping quality
Michel: What is interesting, where should we go?
... Matthias and Christi are interested in animal models
Matthias: Instantiate TMO w drugs from basic research, carried into clinical trials, neurological diseases
... want to connect w human disease; Don't have the manpower to make all the connections
Christi: Trying to solve problem of where sources are and how to get information
... looking at going though the genome, but weren't sure about which approach to take
Zebrafish, can see the organs quickly if you have a disease phenotype
Matthias: Neurological diseases are more difficult, more interesting
Christi: Pharma wouldn't typically use zebrafish, but need to have animal models that will predict safety
... are we seeing what we need to see before we go into Phase I w humans
Joanne: Safety and efficiacy are the big two.
Christi: Don't know of any data sources that would be publicly available.
Scott: Animal models, zebrafish are different from clinical trials (didn't quite follow argument)
Christi: Need predictors for certain types of treatments
Scott: In Netherlands we might be able to get some data on safety and efficacy
... might get into DHL(?) mutations
... VHL above! VHL gives an assortment of kidney disease and other organ function tests
Christi: Looking at use cases on wiki
Joanne: Animal models were strong predictors for norephinephrine use, etc
... analysis at the higher level w diff equations
... order of treatment improvement can be addressed at this high level
ericP how empowering
... order of symptom improvement, above
... had some trouble getting data, but inside hospital it was possible
... bring in a collaborator to give us data?
Matthias: Would be interested in animal models for depression
Michel: Literature around depression, never made linkages to specific mutations
... article talks about some association w. chemical
... curated a pile of papers to pull out dosages, number of people in trial, etc
Joanne: Can we build something that would predict efficacy of a treatment, to be validated later?
Michel: What attributes for predicition. Lot of studies make associations, correlations.
... often recorded in these papers; but results are reported heterogeneously
... perhaps do curated mark-up of papers
Michel: Depression is a nice model, exactly like you suggested
... lots of studies and information about depression, lots of genomics, other omics on depression
... depression covers everything; pharmacogenomics for activating pathways (not getting it here)
... access to patiient data still a problem
Joanne: Can see about contacting my collaborators
Scott: Adrian Coulet used PharmKB to find relations in text mining to RDF
... put on NCBO sparql endpoint
... Anja might be interested over at BioRDF
... Adrian's work based on a favorite set of genes; would be able to limit scope eg "alzheimer's" lexically
Michel: Adopt depression as a focus? Those who have expressed interest in the modalities, maybe think about how relates to depression.
Christi: We have already started around Alzheimer's
... would be neat to move toward clinical side w. depression
Joanne: +1 show something from the patient persective
Scott: Resources? do we have similar materials for depression?
re: Anja might be interested over at BioRDF -> I was saying that I brought it up as a possible resource in the BioRDF work on the microarray use case. Anja wasn't involved.
Christi: We have patient data for Alzheimer's, maybe complete the picture toward depression
mscottm notes that EricP is now talking
Matthias: We did not create too many Alz-specific records
Joanne: Need to do a gap analysis, maybe just a sparql query, from Eric
Eric: Would be happy to continue w Alzheimer's
Christi: +1 to continue w Alzheimer's
Maybe talk w. Tim Clark?
Michel: OK w. continuing w Alzheimer's. These are the two powerful motivating cases.
topic: TMO patient records
Scott: Know people at Leiden w MRI to predict Alzheimer's
... have an idea, maybe contact psychologists for depression