SV_MEETING_TITLE -- 17 Nov 2011

<michel> ScribeNick: bobP

<mscottm2> can't hear you - Michel?

<BobF> severely breaking up

<matthias_samwald> this was noisy AND breaking up.

scribenick bobP

Matthias: Michel and I extended TMO, targeted to represent...

(Michel is talking from an advanced com center)

<mscottm2> LOL

Matthias: after a lot of consideration, edited classes for genetic variantion
... also edited other classes of a general nature
... look at TMO as it is on BioPortal
... namespace of TMO is identical to that used to enhance

<matthias_samwald> http://bioportal.bioontology.org/ontologies/1461

Matthias: so all of changes went live to BioPortal :)
... maybe we should reconsider the authoritative source

Michel: +1 to controlled releases

Matthias: It is going to a release file
... latest on BioPortal 1.02

<michel> http://bioportal.bioontology.org/ontologies/46121?p=terms

bobP sees it

Joanne sees it

Matthias: Tried to follow the realist approach; variation is not an abstract entity like sequence
... this approach tried to build on established principles; represent as material entity

Joanne: Working from authoritative source?

Matthias: Mixture. Looked at DBs and how they representetd variation
... dbSNP also has short indels(?) etc
... still quite open. Is this approach good?
... also, how to represent more detail?
... Ex class region w SNP. Originally from Michel, represent right down to nucleotide level
... SNP would be a subclass of nucleotide (?)
... concerned that this might not be practical. Gets complex beyond simple SNP
... propose to identify SNP thru datatype property, as a sequence stream
... go w. the more normal way that SNPs are represented in real DBs

<Joanne> "as strings of characters"

BobF: Q Current models for variants. SNPs need a reference sequence
... how does that fit here?

Matthias: Refer to loci w/in genome where loci have relation to reference genomes
... found in relation to reference. Would suggest we find ont rep that is independent of standard geneome
... in context of surrounding DNA

<michel> https://docs.google.com/document/d/1BeaU3VJW1S_R3eB5ncrcW9wJ-lMU4xdmxCdjH8E0i5s/edit?hl=en_US

Matthias: but may be useful to keep relation to reference genome

Michel: We did consider where this residue is a SNP, partOf some reference genome
... that's a way to make the connection

<BobF> The NCI Life Sciences DAM may be a useful reference: http://ncientarch.info/sandbox/LSDAM2_2_1/

Matthias: SNP would go out the material branch; reference would go out the information entity branch

Michel: Need to make connection( didn't quite get this)

BobF: Been working w NCI on modeling molecular entities,
... in this model we do address these issues

ericP: Guide us?

<BobF> http://ncientarch.info/sandbox/LSDAM2_2_1/

<ericP> http://ncientarch.info/sandbox/LSDAM2_2_1/

<Joanne> http://ncientarch.info/sandbox/LSDAM2_2_1/

<Lena> @matthias_samwald: if found "reference genomic sequence"@en in the tmo, it this what you meant?

<Joanne> http://ncientarch.info/sandbox/LSDAM2_2_1/

(my image is very narrow!)

Joanne: This works for me +1

Scott: TMO has DNA-region-w-SNP

<ericP> would that look like this in turtle ? [ a NucleicAcidSequenceFeature ] NucleicAcidPhysicalLocation [ a NucleicAcidSequence ] .

Scott: would mke SNP a property of a region

Joanne: SNP needs more than one nucleotide reference

<ericP> or in RDFS: { :NucleicAcidPhysicalLocation rdfs:domain :NucleicAcidSequenceFeature ; rdfs:range :NucleicAcidSequence . }

Matthias: Want to create this ont from scratch on top of TMO?
... analyzed SO a lot; claims to have everything figured out.
... but hard to understand

Joanne: Q (didn't get)

Matthias: SNP needs a population. Use the knowledge to give special meaning to regions of the genome
... liturature tries to avoid mutation (answer to Helen's Q)

Michel: Seq, feature, what are you trying to say?
... actually we have a different molecular material?
... is it really a feature of a sequence?

Matthias: Everything based on individual, w specific regions etc. Then create universals
... to group individuals

BobF: Two different levels of granularity here
... ultimately we are talking about a physically different DNA, should be treating like a physical entity
... from patient's perspective, show me what proceeds from this reality(?)
... different from looking at populations, different perspectives

ericP: Does one of these align more closely? Are use cases similar?

BobP: Both perspectives, but could not drill into this area as much as would have liked
... LSDAM, will be taking these considerations back
... tried to come up w high level use cases, so successful for these
... but instantated in different ways for different use cases

Michel: TMO distinguishes between physical and sequences
... we do have both concepts, both are useful.
... what are individuals composed of?
... we do geno- and phenotyping both

BobF: LSDAM is looking at a translation to OWL

ericP: Opportunity here

Matthias: Could apply a Xform to RDF
... TMO trying to do in this more complicated way, w ont principles
... advantages for reasoning, etc; disadvantage makes things complicated
... putting this realist view on top

BobF: LSDAM used UML b/c of the existing infrastructure
... would welcome the input from OWL experts here

Scott: Taking ont approach so we can use reasoning as well as easy integration
... choice should be tested against a reasoning task
... want a concrete set of test cases, to reason over patient data etc
... would help us ground the conversation

Matthias: Should try to carry this on, try a prototye
... OWL reasoning to classify patients
... practically, different models give confusion on variation in individuals vs in populations

Lena: Patient instance of population?

Matthias: No. A population is a specific entity.

Lena: Patient part of a cohort, so patient could be an instance of cohort class.

BobF: Cohort could be a kind of cohort

(!) kind of popultion

Michel: TMO, would want to assign attributes to populations
... can have cohorts of studies, etc. Averages over individuals, as opposed to specific characteristic

Matthias: Should agree on community process to address this.

Scott: We started w use cases for sparql queries
... Michel put up axioms for ethnicity
... clinical decision support, might not make a difference for which way to go.

Michel: DrugBank geno-pheno all in OWL. Can gain from clear formalization
... mixing categories then we can't leverage from what we have
... want to go beyond just linked data.

ericP: If we write down in LD vs If we write down formally

Lena: Linked data will have new stuff, should be able to express in TMO

Scott: ...different demands

Matthias: TMO approach now is ontological; vs put up raw RDF
... suggest end of Jan 2012

Scott: Pgx paper, there are prods from reviewers
... want more material. Want what we're working on now!
... PGx for Future Medicine

Lena has bragging rights. We await!

<matthias_samwald> try to document all SPARQL endpoints and datasets you create on http://www.w3.org/wiki/HCLSIG/pharmacogenomics_2012 !!

- DRAFT -

SV_MEETING_TITLE

17 Nov 2011

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output