See also: IRC log
<andreasplendiani> +29928aaff is andreasplendiani
<tslater> ted slater is at the 314-274-xxxx number
<andreasplendiani> Hi, sorry I've skipped the "voice presentation", I'm both on call and irc
<matthiassamwald> Hi everyone, I'm sitting in a meeting, so I CAN ONLY LISTEN, not talk.
<tslater> correct spelling of david's name is Nwokeabia
<kidehen> kidehen is "Kingsley_Idehen"
<jzhao> jun zhao
<mscottm> like this: Zakim, number is ircname
<mbevil> ZAKIM 215.652.2134 is mbevil
<kidehen> zakim: +1 617 273 0900 is kidehen
<jzhao> Zakim +0186528aagg is jzhao
<kidehen> zakim: + 781 273 0900 is kidehen
<kidehen> zakim +1 781 273 0900 is kidehen
<AdrianP> Can you please paste the URL to the slides into the IRC
ericP: On slide 1: Drug discovery
ericP: The first question is "Find me genes involved in signal transduction that are related to pyramidal neurons."
scribe: this question demonstrates the value of using semantic web representation
<tslater> ok, Ted Slater is on the phone at the +1.314.274.aabb number, on irc as tslater, but i don't see me showing up in the phone list
scribe: if you attempt to ask the
question via google, you get several irrelevant results
... on slide 4 (we ask this question of pubmed)
ericP: the problem is that the question is still not structured. Now what if we want to ask a structured query against a database with the relevant data
scribe: the goal is to integrate
the databases (go to slide 6)
... on slide 7: the goal is to integrate the data sources such
that the query consults the 4 relevant databases (journal
articles, p. neurons, signal transduction and their components,
...)
ericP: on slide8: we reduce the information to triples and store them in a DB as triples
on slide 9: we see examples of results (that were vetted)
ericP: see slide 10: We began with a 'good-enough model': simple, relationships and classifications
ericP: unification of key terms was crucial for answering the questions
scribe: there were 2 W3C notes that were the result of this work
-> http://www.w3.org/2001/sw/hcls/notes/kb/ A Prototype Knowledge Base for the Life Sciences
-> http://www.w3.org/TR/hcls-senselab/ Experiences with the conversion of SenseLab databases to RDF/OWL
kei: quick modeling was done using RDF, how does this differ from using OWL?
ericP: The data is writen in RDF, some of it is described in OWL, however, the OWL modeling was not done upfront
ericP: the Gene ontology is an example of where OWL was used to describe classes of protein
<ericP> http://tinyurl.com/55ozfp
vipul: the methodology has been to function in a consensus manner
scribe: there was a significant
social engineering component
... the driving force is consensus not technology
vipul: the first decision was the usecase to develop for the prototype
scribe: various usecases were discussed and the decision was made to focus on patient recruitment
vipul: a main assumptin was that the data was in an EMR in order to demonstrate interoperability
vipul: how do we reuse clinical data for research?
scribe: HL7 RIM can be viewed as
a format for clinical practice data
... CDISC can be viewed as a format for clinical research
questions
vipul: suppose we had a mapping module, how can it be made flexible to encorporate changes in these formats
vipul: there was a lot of debate with no answers
scribe: we used GALEN, CPR,
helen's eligibility criteria ontology, SDTM, HL7 RIM,
etc..
... the conclusion was that no one ontology or information
model was likely to fit the bill
... so the goal was to align with existing terminologies but
address inadequacies in these source terminologies
vipul: the goal is to provide constructive feedback to the originating standards bodies
vipul: the goal of extensibility is still a target
vipul: we want to demonstrate re-use
scribe: patient recruitment
against EMR is a demonstration of re-use of clinical data for
research questions
... we want to show that even though data is coded via
SNOMED-CT how can queries expressed in NCI thesaurus terms be
invoked on the source data to answer questions?
vipul: two of the seed models: SDTM & HL7 RIM
vipul: we are mostly interested in re-using the mapping for other scenarios.
scribe: for example: consider
hl7:substanceAdminstration and
sdtm:concommitantMedication
... we would like to reuse these terms across various
scenarios
vipul: decision 3: this was a key decision - We implemented teh POC using a real world dataset instead of a synthesized dataset
scribe: what would be an
appropriate seed information model for the data we use?
... We are very thankful to Parsai for providing the source
data
vipul: the information model mapping is the core of the effort
kei: is the seed data publicaly available?
vipul: in the POC, the data is anonymized
scribe: there is some demographic
data, however
... in the real world, we will need to address privacy and
encryption issues
... we are transforming the queries and not the source data
formats
<ericP> we are pushing the query all the way into the database
vipul: the W3C is clearly interested in working with external standards organizations
scribe: there is an implication
about w3C being content neutral
... this doesn't mean that gaps/inadequacies will be
represented as is
... the goal is to address these inadequecies and give feedback
to the standard bodies
vipul: Bron Kisler (from CDISC) is interested in such a collaboration given our content neutrality
vipul: (the architecture) the mapping module transforms queries (in two places) in order to align the queries with the source data
vipul: ericP is experimenting with SPASQL query engine for purposes of the POC implementation
vipul: we made an assumption that terminology in clinical trials and clinical practice are different
scribe: the technology choice is: SPARQL and N3 rules (we prefer N3 rules)
vipul: there are issues with the mapping between concepts related to medication and medication classifications
scribe: the usecase is such that
the criteria is specified, this is translated to HL7 RIM, this
is translated to the underlying SQL-like language, then
component 6 identifies patients relevant to the query
... there is an open question about the expressivity of SPARQL
vs. N3 rules
<AdrianP> declarative rules vs. SPARQL, ... e.g. no real rule chaining in SPARQL
<vipul> http://esw.w3.org/topic/HCLS/ClinicalObservationsInteroperability/SPARQL8.html
ericP: we have some work to do yet on the query translation component
vipul: a couple of months remain on this task
<vipul> http://esw.w3.org/topic/HCLS/ClinicalObservationsInteroperability/InterOntologyMapping.html
ericP: Translational medicine is important for determining which medications are relevant at the point of care based on the evidence of controlled research
scribe: the task is rather similar to computer-assisted diagnoses
<matthiassamwald> what you describe sounds more like "personalized medicine".
ericP: the interest in Pharmaceutical Companies is how to improve the effectiveness of their medications from the research done
mscotm: okay
mscottm: translation medicine is more 'bench-to-bedside' where we are attempting to connect two disparate areas: clinical practice / data and biological research
scribe: personalized medicine is one application within this larger realm
mscottm: ericP was interested in using the wiki to solicit feedback on other projects to consider
scribe: esp. now we have new
members with new datasets and new problems relevant to this
interest group
... is there anyone out there working on a particular project
relevant to translation to RDF/OWL?
<AdrianP> http://esw.w3.org/topic/HCLSIG/Project_Ideas
mscottm: there was a recent project suggestion we can use as a strawman
<ericP> http://www.w3.org/mid/486217F4.6020205@inf.unibz.it
<mscottm> http://www.w3.org/mid/486217F4.6020205@inf.unibz.it
<andreasplendiani> nope
<andreasplendiani> access restricted
mscottm: (reads out the idea sent to member-restricted archive)
scribe: Reading through the meeting minutes, KB description, and proposals for
the KB enhancements, I was wondering if you think there may be interest
in extending the technology-part with "ontology-based data access"
ericP: adding the project to the Project_Ideas page
mscottm: This would be a project under the KB enhancement section
ack, vipul
vipul: if we suggest technology extensions we should include a description of the ROI
<mscottm> vipul: when suggesting a particular technology for a project, one should explain the value of that particular tech to the user
<ericP> http://esw.w3.org/topic/HCLSIG/Project_Ideas
<ericP> KB++ OWL Ontology queries over RDBs
-> http://esw.w3.org/topic/HCLSIG/Project_Ideas/OWL-RDB "OWL-RDB"
This latter link is where we are filling out the project idea
kei: perhaps we need categories of projects?
scribe: other categories: terminological frameworks, use-case driven (specific to domains), drug discovery, diseases, traditional chinese medicine
<marco> I am afraid I have to leave you. Thank you for the interesting meeting. I will disconnect silently to not disturb too much. Goodbye!
<mscottm> bye marco
marco: take care
<ericP> http://esw.w3.org/topic/HCLSIG/Project_Ideas/OWL-RDB
One value proposition / ROI of this project idea could be the ability to reuse organized, existing ontologies as a way to query more idiosyncratic data sources
scribe: this would allow researchers to query the KB using consensus terminology
kei: perhaps we need footnotes, etc. to better explain the projects
AdrianP: are we moving to Semantic Media Wiki?
ericP: yes we are
<AdrianP> great
<kidehen> ericP: do ping me re. Semantic MediaWiki etc. in relation to Virtuoso hosting etc.. when you have a moment
kei: one proposal: Extending the senselab project to add more content to existing data
scribe: considering a
translational medicine component
... linking this to the clinical side
<zenhack> Marc Hadfield / nic zenhack / 212-217-9956
<mbevil> Zackim, +1.215.652.aaee is me
<zenhack> zakim +1.212.217.aamm isme
<kidehen> zakim +1 781 273 0900 is me
<kidehen> zakim +1781273.aahh is me
<AdrianP> bye
<kidehen> Zakim who is here
<andreasplendiani> bye
<chimezie> I unfortunately, have a hard-cutt off @ 12:30 EST PM and thus will have to sign off
<mscottm> bye and thanks chimezie
np, take care :)
<mscottm> ericP: it is useful to use membership list to manage the mailing list BUT it remains open for public comment. So, non-members can still request subscription by sending mail to eric Prud'hommeaux <eric@w3.org>
<mscottm> Also, we are open to Invited Expert applications
<mscottm> you can apply for Invited Expert status here: http://www.w3.org/2002/09/wbs/1/ieapp/
<mscottm> bye all