See also: IRC log
michel: new directions to solution problems
rich: working on clinical
proposal - product labeling; solidify the resource, making it
useful for NLP folks. links back to the LOD
... drug interactions and pharmacogenomics; colleagues in
school of pharmacy, pgx guidelines of interest;
pharmacy-specific use cases
... going to brainstorm specific tasks in next meeting
... modeling the semantics of pgx claims in product
labels
... have a semantic model to capture 29 statements from product
labels
michel: sounds really good - looking forward to the requirements, and determine how we can support or collaborate in this work
matthias: what semantic model are you referring to?
rich: met with clinical pharmacology and pharmacy people to pull out pgx sections and developed a model rather organically
bob: working with rich for a few
months to structure some of the product labels, in a way that
we can access and mine
... will be really useful with the linked data clouds, like
pharmgkb
<matthias_samwald> lots of typing noise!
<boycerd> leaving now, happy thanksgiving.
bye rich!
bob: amia group - significant interest in clinical pharmacogenomics. overlap - could grow our ranks to advertise to that channel
michel +1
matthias: the additional focus will help to bring more people to the table
bob: graduate students
matthias; one goal should be to align our interests
rich: work that matthias presented on 2d bar codes - really neat application. would like to consider how to apply this more broadly to other knowledge sources
sorry - bob said the above
bob: hla application is ripe for
this - number of loci should be containable in the bar code,
but extreme variety in the population
... blood typing?
... would like to think of additional ways in which we can
coordinate that
paul courtney - dana farber cancer institute - blood and marrow transplant group there; hla typing is really important. how we match donors with recipients.
matthias: definitely applicable
bob: invovled in national bone marrow donor program
matthias: a standard - shared ontology, shared design pattern - for representing pgx / immunogenetic data
lost my connection
<matthias_samwald> me too
<matthias_samwald> back on again
bob: structured product labels -
the pgx section is a big paragraph of (mixed) text; knowledge
is completely unstructured. e.g authors can say whatever they
want about gene interactions - just an association with a gene
and drug, or it could be a specific allele
... how can we mark this up so they can be mined later
... rich has a tool to extract triples out of the markup;
however, they vary enormously about what they say
matthias: manual annotation is really required
bob: +1
... at the point where we can take the resource and find
associations between genes and drugs; but more detailed
evidence of the nature of the interaction is not guaranteed
paul: is it that pharma just doesn't have access to that information to provide more specific details?
bob: personal opinion - not an
unwillingness to share the information - but it's up to the
label author to make an assertion without all the
evidence
... don't want to make an incorrect assertion - it may be more
prudent to state the association
michel: cds ontology
matthias: genomic cds ontology
http://www.genomic-cds.org/ont/genomic-cds.owl
... demo with patient data: http://www.genomic-cds.org/ont/genomic-cds-demo.owl
... extracted data from dbsnp and pharmgkb, ontology that
contains the snps and logical definitions
... clinical guidelines are represented - 50 separate rules to
match specific patient profile with specific clinical
guidelines
... some guidelines refer to snps, alleles and genotypes
... with the ontology, can infer one from the other
... open question on how to represent phasing rules in the
ontology; ambiguous combination of snps;
... missing representation of copy number variations
... ontology assumes one copy of each allele on each
strand
... CNV have significant clinical impact
... more information about genetic assays for testing specific
alleles
... process for annotating the rules and extracting the data
need to be automated; better provenance
... in the email, definitions provided in owl manchester
syntax; possible to have a sentence and generate an axiom
bob: great work !
michel: +1
matthias: mostly script
generated. uses dbsnp service to get more information
... uses 2 excel spreadsheets - allele/haplotype definitions
from pharmgkb; clinical decision support rules that were
manually curated
... spreadsheets make it possible for domain experts to
validate
... move this to google doc and collaborate
michel: michel asks bob what the status of the clinical guidelines are
bob: developed a information
model for the guidelines, currently evaluating it
... hope to have one of those CPIC guidelines in a structured
format
... by the end of the year
... captures detailed information about the guidelines -
includes link outs
... as specific and necessary to capture that important
information, but general enough to capture numerous different
guidelines
... just wants to convince himself about the approach
matthias: drug + genetic markers + evidence - just basic info
bob: guideline with hlad
genotype; needs revision; until i get to fix this, i want to
keep working on it, but will present asap
... HL7 clinical genomics group?
matthias: conflicting / heterogeneous / overlapping projects ;
ericP: CDA (xml representation of v3 (RIM) concepts), FHIR (based on v3); similar, but a number of false equivalences between these. RIM is like an upper level ontology
bob: use cases that hl7 clinical genomics could be used for us
michel/ericp: in theory we could use FHIR data to translate into genomic-cds
This is scribe.perl Revision: 1.137 of Date: 2012/09/20 20:19:01 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) No ScribeNick specified. Guessing ScribeNick: michel Inferring Scribes: michel WARNING: No "Topic:" lines found. WARNING: No "Present: ... " found! Possibly Present: achille_z bob boycerd ericP matthias matthias_samwald michel paul pkc_nh rich You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy Got date from IRC log name: 21 Nov 2012 Guessing minutes URL: http://www.w3.org/2012/11/21-HCLS-minutes.html People with action items: WARNING: No "Topic: ..." lines found! Resulting HTML may have an empty (invalid) <ol>...</ol>. Explanation: "Topic: ..." lines are used to indicate the start of new discussion topics or agenda items, such as: <dbooth> Topic: Review of Amy's report[End of scribe.perl diagnostic output]