HCLSIG/PharmaOntology/Meetings/2009-12-10 Conference Call
Conference Details
- Date of Call: Thursday December 10 2009
- Time of Call: 12:00am - 1:00pm ET
- 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: 42572 ("HCLS2").
- IRC Channel: irc.w3.org port 6665 channel #HCLS2 (see W3C IRC page for details, or see Web IRC)
- Mibbit instructions: go to http://www.mibbit.com/chat and click the server link. Enter irc.w3.org:6665 into that box, enter a nickname, and enter #HCLS2 for the channel
- Duration: 1h
- Convener: Susie
Agenda
- Paper Submission - Susie
- Ontology Update - Colin
- Next Steps - All
- AOB
Minutes
Attendees: Chris, Joshua, Michel, Christi, Lee, Joanne, Elgar, Trish, Susie Apologies: Bosse, Colin
Susie - Paper submitted, great material, Michel did a great job of getting the data in a triple store
Christi - when will we hear?
Susie - not sure, mid-Jan? Thank you everyone for all your work. Wonderful example of group with different backgrounds generating more than the individuals can provide
ONTOLOGY UPDATE Susie - Colin sent out emails, worked with Michel to add new terms as defined classes, talking to Luke with ACGT ontology and comparing notes. Also sent email on orphan terms.
Michel - major problem is that we are not sure what data will be tagged/annotated. very useful to have DX criteria and know how to have information. Pretty sure that we will have toxicity profiles. Really good ontologies are driven by the data to have the necessary terms
Susie - will you be following up on this with Colin?
Michel - if we come to consensus then we can have others review. Also working on better utilization for mappings to better query all data in triple store.
Susie - was this because there was a tight time-frame or does additional mapping effort needed.
Michel - the mapping came at the end. if chance to review paper then we can add this info with TMO terms.
Susie - yes, this was a concern but time was tight
Michel - had data from different data sets with different formats, this heterogeneity across data sets makes it hard to query - need to really understand the data set.
Susie - even if querying a linked data set, need to understand the data to run the query and compounded significantly with more data sets
Joanne - really need to know the structure and different semantics amongst the data sets
Christi - jot down observations and problems that occur
Joshua - seems like one of the selling points of using these representations is the pay as you go integration.
Joanne - public data sets, able to serve as template to proprietary data in-house to better map to internal data sources
Susie - good point since ontology would be used in-house. challenge is gaining access to proprietary data. map data to ontology.
Bosse - from Pfizer, have been asked to get access to HIV data. will be collaborating with GSK and will show need for standards in this space. how to prevent people from making incorrect associations.
Susie - concerned about policy of the links within the public linked data set, not necessarily saying that they are not great but that people may not understand the data.
Christi - data is stored in so many data sets that it is not linked correctly
Susie - place for pre-competitive consortium to pull together data to ensure links are proper
Lee - Pfizer wants to do more outreach. also doing a lot of vocabulary work, but very early work. W3C will be a really key group in this
Susie - interested in these opportunities for synergy. W3C interested in creating standards at low-level, there can be collaborative efforts where final work can be submitted to W3C for stamp that standard is good.
Lee - continue to facilitate this, would prefer more that companies get together on what they need. will synthesize this for potential next steps.
Susie - continue on use cases that relate to Pharma companies, would also like to continue to pursue the physicians role
Christi - agree, that to be translational needs to span these areas and this compartmentalization is what has happened in the industry
Susie - want to make sure that the ontology is optimally designed where pharma can use all integrated data
Susie - maybe split the discussion into 2 - physician and pharma - all with same ontology and body of knowledge
Susie - Physician scenario further work for user interface, Joshua is interested in looking at interfaces for EHR
Joshua - caBIG, cancer bioinformatics grid funded by National Cancer Institute. would like to build out open source EHR system customized for oncology and deploy to cancer centers. looking at semantic web technology that can fulfill these specifications, determine trial eligibility. Goal is to make trial matching more efficient.
Susie - this would be primarily for trial matching, work for EHR, who would be the user for trial matching software?
Joshua - produce working app to be used by physicians. Identify functional modules as general purpose services. one module would be a trial matching module. both physicians and patients would use this.
Susie - what sort of technology would is being reviewed?
Joshua - all technology coming out of HL7 and caBIG is service oriented and has web service SOAP based implementations. For semantic web tech, if we can take a large set of eligibility criteria (Aspira?? project) can take these to describe relationships. just some initial brainstorming, not sure how TMO can be used.
Susie - not sure if there is an answer yet. thinking of an interface with broad utility similar to VA system in US (liked by physicians) would help with with diagnostic. Thoughts from the group?
Christi - physician needs to be able to source the trial quickly, from pharma need to find people for the trial.
Michel - in working with the data, until there is a categorical system it will be difficult to work with this.
impressed with personal health records, but missing linkage from description of drug
Susie - following up, with clinicaltrials.gov linked data, much harder to find info since inclusion and exclusion data is in free text
Michel - do these other systems specify these terms
Susie - not aware of this
Michel - if there are no source vocabularies then we may need to add this.
Trish - Annotator of clinical information
Michel - realized that phrase needs to be represented
Susie - Michel can you see if these systems exist? i can check with people that write these reports to see what the source is
Joshua - can ask Aspira?? group for this information. also work from Stanford (ERGO)
Lee - to bring back to pharma, really important to have this
Susie - seems interesting avenue to pursue. the other area to pursue was data sets for drugs. mapping RxNorm?
Michel - if we can get data for this then that will help
Susie - can the national drug codes be RDFized?
Christi - these are package specific, depends on manufacturer, type of delivery, dose
Susie - are there different names for drug, delivery mode, etc?
Joanne - has ingredient, delivery mode (injection, pill), brand name, drug pack
Susie - Think of next papers - BioOntol SIG, focus on research side of things
Chris - UPenn yearly conference from Gerald Fitzgerald, although think that this is by invitation only, will follow-up
Christi - any news from TBI
Trish - no not yet
Susie - not aware of other groups that have heard
Elgar - CHALS presentation on TMO
Susie - would be good to coordinate on this since there will be W3C presentations also