See also: IRC log
Notes begin here
<sudeshna_> http://esw.w3.org/HCLSIG/SWANSIOC/Actions/SWANmyExpArray#Ontology
SD: Would you all please go to the above link, 2nd draft model
... classes in purple are existing SWAN classes; pink are existing myExp classes; orange are proposed new ones
... Study is main class; typical research workflow - hypothesis -> study -> experiment; for Study we can import from OBI as subclass
... Study motivated by Res Question and has Hypothesis, Design (e.g. timesoucrse, retroactive, etc.)
... Study designs could be taken directly from OBI
... Study has Type - this is a convenience to know what we are dealing with, this is a controlled vocabularly of types of studies; new class
... Factors are the experimental variables - chem, bio, time, etc. - overlap with Exper Factor ontology dev at EBI by H Parkinson's group; would like to map to them and use their factors
... Study also has DataAcquisition = 1:1 correspondence with OBI class
... Data Acq can be done by measurements = ht, wt, MRI, etc etc; or could be done by an Assay, i.e test carrying out in the Lab - part of OBI so import
Assay implies BioMaterial which has characterisits
SD: Exp Factor Onotlogy we can use directly, Helen Parkinson will be in Bosont for ISMB, we can discuss further then
... Assay produces SWAN PrimaryData; ComputedData is a result of analysis; results in a Claim
... we have a very simple version of myExperiment at a high level - can David help us flesh it out?
... Use cases: (a) gene list from BioRDF; (b) Tim & I have been working on e.g. genetic association use case in Autism, e.g. GWAS; as next step let's take some of those studies and see how that use case maps out to the model
... again, David can we get some help on this fleshing out?
DN: some of the myExp fleshin gout is in the services and services provenance
SD: perhaps as action item we can work offline to take a particular genetic association study and model it here
DN: can provide some diagrams for the next call
SM: can DN confuse us just a little by saying what the sticky bits are? DN: what sticky bits? SM: what level do the models articulate? is there a big problem here?
DN: analyses done without myExp?
SM: how do we glue all this stuff together?
... as myExp is so much about services?
... for example do you have stuff in myExp about GWAS?
DN: essentially workflow made up of interlinked processs - some effort to express what type each is - but limited - the "services" are more about WSDL, SOAP, etc - can we use BioCatalog here to classify the actual services themselves?
... hoepfully we there get the details on what services are really doing
Satya: how does description of service usable in contet of describing how an experiment is being conducted?
DN: ???
Satya: in provenance community we have looked at limitations of services; what we need in context of experiment is in semantic annotations; more covered by OBI than by a service oriented ontology
... what I'd look at is extend Taverna workflow model to add semantic annotations, extending Provenir - for these use cases may comncentrate more on domain aspects of these anayses
<ssahoo2> Provenir - an upper-level provenance ontology: http://wiki.knoesis.org/index.php/Provenir_Ontology
SD: we really lookijg at e.g. the kind of statistical analysis etc
... agree with Satya
SM: use case of descriinb experiments as workflows in the sense of "computable pieces & types of statistical analysis" but graph of workflow can allow you to find similar processes
SD: need to capture summary that susch and such statistical method was used
Satya: looking at computational processes - e.g. sequence extraction, stats analysis, etc
<mscottm> Autism was mentioned earlier - I've noticed that autism is being studied by NCBO: http://bmir.stanford.edu/publications/view.php/ontology_driven_data_integration_for_autism_research
DN: BioCatalog is based in manchester as well, like Satya's collaborators in Manchester
Satya: I interacted with the BioCatalog interface thing too
SD: what was the name of the effort, Satya?
<ssahoo2> Janus semantic provenance framework
SM extension of Provenir
SD: good discussion
... ACTION ITEM: work to define one particular use case to capture details of work flow for next call;
... Satya - can you update us on Microarray use case
Satya: The model you shared from Paolo - has many overlapps with OBI - lets use where that occurs - SD: can you show us the specific classes where you see overlap?
<ssahoo2> http://esw.w3.org/HCLSIG_BioRDF_Subgroup/MicroarrayExperimentContext
SSatya: posted this - use case for micrarray use case from BioRDF
Satya: what we have done is primarily (Kei did this by hand) created scenario of this gene list as RDF graph, Jun and Satya created a schema taking concepts from OBI and other concepts at NCBO - lena has integrated with the DiseaseOme
<ssahoo2> http://ibl.mdanderson.org/~mhdeus/sparql_federation/endpoint.php
<ssahoo2> diseasome dataset
Satya: this is the diseaseOme endpoint
... can find a list of diesease associated with this gene
SD: looks like m-array use case you are working on in BioRDF group - lot of overlap - we can learn from each other - rather than direct use of each other's work -
Satya: significant overlap but we didn't deal with Study, Rsearch Question, etc -= these are probably sepcific to scientific discourse
SD: action items: take on of the use cases & map onto the ontology - work out the workflow with correct semantic annotations and meets scientific needs
... can DN help - ? DN: can produce some diagrams for workflows - have a lot of other stuff going on - for next few weeks I'lldo some diagrams
... four weeks out
AG: question: everything hooks from myExp to computation - if I am doing insilico biology - everything under computation belongs to myExp - can this be something jsut run on my computer? shouldn't it be considered as a computational assay?
AG, SN & TC: discussion on whether this is true, whether there is such a thing as a computational Assay, is it the same as a physical laboratory Assay?
AG: whay should a computational experiment be different form a lab experiment?
Satya: what distinguishes in silico from in vivo? point is how well does the computational model mimic the natural world?
Paolo: OBI says exactly what TC said; Assay produces data from real world; DataTransformationis data to data
SM: what we're talking about is what you call something; point remains that there are processes that are going on here choices being made about storage of enzymes, conditions, temperatures, laser config, etc -- conutless factors have an effect - same holds for computational processes
<sudeshna_> Let's make a use case and see how it plays out instead of changing the model directly
TC: let's maintain the distinction between inductive and deductive
... let's have the people interested in computational experiments do a specific use case here
AG: fine with me - I will look at OBI
SM: this call shouldn't have four weeks intervals