<AndreaS> Hi

<mscottm> hi

<scribe> scribenick: matthias_samwald

janos: started when i looked into a database called Biogrid

<mscottm> Andrea: briefly: involved in BioPax and semantic web aspects of BioPax

janos: interesting, curated database
... protein-protein interactions from many organisms
... i downloaded a dataset, loaded it into MySQL
... i saw potential for linking it to LODD -- pubmed IDs, gene ids etc.

<mscottm> janos - your voice is dropping out for me. is it only for me?

janos: searching for 'biogrid' and 'owl' gave some results in google
... pathway commons, for example. they have Biogrid in BioPAX format.
... so it was available in OWL, but not as linked data.
... i downloaded the BioPAX OWL file, loaded it into triple store

<ericP> -> http://www.w3.org/wiki/images/3/30/BioGrid2RDF.pdf Experiences with http://thebiogrid.org/

janos: qustion is: can we take something in this format an link it to other resources on the Semantic Web?
... i chose to work with the tabular raw data
... for example, with PubMed ID you can easily link to Bio2RDF URIs
... slide 8: for example, you can put a gene name into NCBO annotator, it will correctly give you the gene ID

scott: i have a question. it looks interesting for setting up a script that does that for each item in the column to retrieve PURLs. could you do that?

janos: can be done
... i am now working on a version of UMLS that includes CUIs as well as other identifiers
... can facilitate alignment
... Bio2RDF also has Entrez Gene IDs, this is another opportunity for linking
... on the final slide: an example of a detail in the UMLS metatheasaurius
... you can see that is also has PubMed IDs, but not linked to external resources
... the project is in an early phase.

any questions?

scott: the interactions are between what types of entities?

janos: data comes from high-throughput screening of protein interactions
... protein is coded via gene, but data is about protein interactions
... data is manually curated
... i am also working on a project that is part of VIVO
... could also be used to connect researchers working on strongly related proteins
... i don't want to re-do what Pathway Commons did. Rather, I annotate the data that they already have.

andrea: could you identify any limitations of the approach? you also worked with the tabbed format -- why that instead of BioPAX?

janos: i have not done an exact comparison between the two. I suspected that the different formats did not contain exactly the same version of the data.

andrea: did you check the consistency of the BioPAX file?

janos: no

andrea: UMLS is not completely free, can you directly use it?

janos: they have different levels of restrictions, i only used level 0 (freely available) subsets of the UMlS
... as long as it is level 0 you can publish it as RDF

eric: if i map something from SNOMED CT to something else, i am not allowed to recapitulate it... but can i use the terms from SNOMED CT in my alignment at all?

janos: if you only use the identifiers and not the original wording of the descriptions... there might be a problem

scott: john madden believed that it was no problem to use the terms. the problem is when you make the entirety of SNOMED available for download on your website.
... for the Shared Names initiative i wanted to be able to serve PURLs for SNOMED terms
... what we came away with was that it should be possible to server the PURLs/identifiers, but you should not make it possible to download the entire SNOMED vocabulary from your site

claus: i am looking at the BioGrid web page. data is available in many formats. maybe we could talk to them to serve RDF as well?

janos: i have not contacted them yet. what i wanted to do so far is to add the annotations to the web of linked data. but it could be good to talk to them.

scott: lets get back to the topic of the W3C note
... if we were to produce a not with a step-by-step guide for publishing linked data in the life science domain, which timeline would we be looking at?

eric: the social process is hard to predict.
... figuring out how to reach consensus

scott: i was suggesting that we could take material from the best practices document we recently submitted, revise and reformat it -- then we could see how it looks and decide how to proceed further. This could be done rather easily.

eric: the process for a recommendation would also include peer review, reference implementations. this would take one year at least.

will join the call in a minute...

<michel> dd

<mscottm> congrats

<mscottm> nope

<Bob> scribenick Bob

<michel> scott: collabrx - mike travers no longer working there, but is available to chat

<mscottm> Jeff Shrager

<michel> .. AI guy

<Bob> Scott: Jeff is LISP/AI guy

<Bob> ... :-) have internalized the LOD cloud

<Bob> ... converatons wiki-style about patients, alleles

<Bob> ... We could try for 2 weeks

<Bob> ... Adrian Coulet on the 14th

<Bob> Michel: Will try to get Jeff in for the 21st

<Bob> ... 21st is Harmony meeting NYC. I'll be attending

<michel> https://docs.google.com/document/d/1lKdDSb2uBBIeTEQAv2CyTHN_aVW63k9si1hmmilOMi0/edit?hl=en&authkey=CJGUtcwF#

<Bob> Michel: Paper. Matthias has looked thru instructions for authors

<Bob> Matthias: J. Fut. Med.? research article, create our own ontologies or focus on...

<Bob> Michel: Scott invited to contribute to J.Pharmacogenomics, initially for review paper

<Bob> ... advanges/disadvantages of different approaches, ultimately get to ontology

<Bob> ... consider conceptualization in ontological approach

<Bob> ... May 3 time frame

<Bob> Scott: skype conversation here:

<Bob> ... sopharm, nanopubs, review of technologies and how they have been used to date

<mscottm> Review of representation in SO-Pharm, PGx knowledge, Nanopub, anonymous

<mscottm> Pharma co., relational in general, tables to RDF and OWL, identifiers, + proposal

<mscottm> review of representation of PGx knowledge

<mscottm> from relational database to logic-based representations#

<mscottm> identifiers, terminology, database cross-references

<mscottm> entity-relation models

<michel> https://docs.google.com/present/edit?id=0AYy0zfdRviKsZGM5cWM1OXRfMTAwZzlycmJjZmI&hl=en&authkey=CKXAsvMI

<Bob> Scott: From last week, discussion of how we want to represent real data

<Bob> Mihel: We have specific use cases to draw from

<Bob> Michel: We need descriptions of resources; three use cases, one of which is diagnostics

<Bob> ... another use case for clinical trials, id genetic cohorts, side effects

<Bob> ... 3rd use case research

<Bob> Joanne: Diagnostics and treatment not nec coupled

<Bob> Matthias: Narrow down on pharmacogenomics, want to focus

<Bob> Joanne: Don't want to lose sight of the bigger picture.

<Bob> Scott: Idea of using molecular characterization for stratification of patient

<Bob> ... use semweb rep to do patient stratification

<Bob> ... tangible form of example for everybody

<Jluciano> (I find it amusing to hear that patitent startification is the new buzz --- I presented a way to use Nerual Nets to do this in Sept 1994 at an Irish Neural Net conference.

<Bob> Michel: Patient-centrict is most appropriate, add to TMO work

<Bob> Joanne: Data from IO Informatics, yes it has genetic data

<Bob> ... genes, biomarkers, blood

<Bob> Michel: But link to PGx outcome?

<mscottm> http://www.genomeweb.com/dxpgx/systems-biology-fights-cancer

<Bob> ... HCLS wiki page for translational medicine, links to FDA alerts

<Bob> ... time frame short for IO data, maybe later

<Bob> ... need someone to develop this patient-centric use case

<Bob> (got the link?)

here is the list of FDA drugs with pharmacogenomic-relevant info: http://www.fda.gov/Drugs/ScienceResearch/ResearchAreas/Pharmacogenetics/ucm083378.htm

<Jluciano> who's voice is that?

<michel> http://www.w3.org/wiki/HCLSIG/PharmaOntology/UseCases

<mscottm> Bosse

<Jluciano> Thanks!

<Bob> Bosse: Chris had a lot of use cases, including PGx

<Bob> Michel: We had developed a use case; maybe look into this one!

<Bob> ... need validation on treatment option, etc

<Bob> Scott: I have contact info from founder of Co. making chemo chip, using biomarkers

<Bob> ... can contact radiologist in NL w. data

<Bob> ... siimilar to OncoDx

<Bob> Joanne: We can get warfarin data, synthetic?

<Bob> ... probably available today

<Bob> Scott: Can we follow the whole (life cycle) of a (patient encounter)?

<Bob> OK: Not whole scenario! Just do you have genomic info, sufficient to stratify

<Bob> (I overstated Scott's proposal!)

<Jluciano> Here appears to be clinical avatar data: http://lpm.hms.harvard.edu/sites/default/files/cairodemodata.tar_.gz

<Jluciano> the clinical avatar data is on warfarin

<Bob> Matthias: Draft OWL ont for treatment decisions

<Bob> ... separate ontology, maybe could be integrated w. what we need for PGx

<Jluciano> from Peter Tonellato's group at Harvard Laboratory for Personalized Medicine.

<Bob> Scott: Represent rules for this? or sparql?

<Bob> Matthias: Looking at SPIN too, haven't fixed on how to pursue decision support

<Bob> (don't let me overstate all this :-)

<Bob> Matthias: Maybe can offer a prototype that could be a start for work

<Bob> Michel: can you hear us?

<mscottm> Matthias: What I could offer now is a prototype that hasn't left the lab

<Bob> probably not :(

<Bob> Scott: What kind of decision support?

<Bob> Matthias: Start, mostly focus on order entry. Medication to recommendation, warnings, etc

<mscottm> What's your number? I'll try to dial you in.

<Bob> ... focussing on the the process of specific medication as it applies to a patient

<Bob> Scott: Not a larger scale disease frame?

<Jluciano> http://clinicalavatars.org/#/menu_model

<Bob> Matthias: Not at the moment. But looking at neurological diseases and cancer.

<Jluciano> http://clinicalavatars.org/#/menu_model/menu_doWhat?model=warfarin

<Bob> Joanne: Breast cancer and warfarin use case in this clinicalavators.org

<Bob> ... can use tools to create datasets

<Jluciano> AGE GENDER RACE HEIGHT WEIGHT SMOKER DVT AMI CYP2C9 VKORC1 >64 F White <58 in <100 lbs N N Y 1/1 B/B >64 F White 60 250 to 259 N Y N 1/1 B/B 25 to 44 M White 66 180 to 189 Y N N 1/1 A/A <18 F White 73 150 to 159 N N N 1/1 B/B <18 F White 64 120 to 129 N N N 1/1 B/B 45 to 64 F White 72 160 to 169 Y N N 1/1 A/B <18 M White <58 in 130 to 139 N N N 1/1 B/B 45 to

<Bob> Joanne describing how to build your own avatar

<Jluciano> http://clinicalavatars.org/?sid=doseDiagnose&preConfig=warfarin&downloadAvatarsFileName=/tmp/AVATAR3fk535

<Bob> Michel: Don't know what headers are, etc

<Bob> Joanne: pasting info now...

<Jluciano> Warfarin is the most widely used anticoagulant in the world. However, its narrow therapeutic window results in a relatively high risk of bleeding when over-dosed or an increased risk of thrombosis when under-dosed. Furthermore, the wide-range of dose response for an individual is heavily influenced by variants in the CYP2C9 and VKORC1 genotypes that control w

<Jluciano> Chih-Lin Chi, Ph.D.1, Prasad Patil, B.S.1, Vincent A. Fusaro, Ph.D.1, Peter J. Kos, M.S.3, Rimma Pivovarov, B.S.1, Charles F. Contant, Ph.D.2, Peter J. Tonellato, Ph.D.1,3 1Center for Biomedical Informatics, Harvard Medical School; 2TIMI Study Group 3School of Public Health, Univ. of WI-Milwaukee

<Jluciano> Our clinical trial simulation design consists of four steps. First, we identify variables from a literature review to create the stochastic model that produces clinical avatars based on statistical characteristics of a population and variable dependencies. Second, we predict an initial dose for each avatar using dosing algorithms from literature. We then appl

<Jluciano> protocols used in clinical trials and other studies as guidelines for daily dose, dose adjustment, and INR measurement schedule. Fourth, we conduct randomized trial simulations on clinical avatars. In our clinical trial simulation, we randomly assign clinical avatars to different treatment protocol groups and compare the TTR between different groups.

<Bob> Michel: Start by creating diagrams, writing descriptive axioms

<Jluciano> INR Response --- International Normalized Ratio (for blood clotting time)

<Bob> ... can make diagrams on Google docs

<Bob> Scott: We can make diagram for INR-warfarin in google doc

<Bob> Michel: Dosage, compostion, diagnostic, outcomes

<Bob> ... extend w. gene-drug relations

<Bob> ... Note that you can also put comments in google doc, w. a little bubble

<Bob> Joanne: Settled on warfarin?

<Bob> Michel: Yes, very well understood.

<Bob> Joanne: we could switch over to breast cancer

<Bob> Scott: Already working on breast cancer, have some resources

<Jluciano> http://clinicalavatars.org/?preConfig=gailRosner&pid=generate&preConfigured=GailRosnerUSMixedCPT&numAvatars=100&submit.x=152&submit.y=35

<Bob> Joanne: Link for breast cancer above

<Bob> Michel: Will create the diagram.

This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found ScribeNick: matthias_samwald
Inferring Scribes: matthias_samwald

WARNING: No "Topic:" lines found.

Default Present: mscottm, +44.453.083.aaaa, +1.631.444.aabb, +1.647.839.aacc, +1.781.431.aadd, +44.754.501.aaee, EricP
Present: mscottm +44.453.083.aaaa +1.631.444.aabb +1.647.839.aacc +1.781.431.aadd +44.754.501.aaee EricP

WARNING: No meeting title found!
You should specify the meeting title like this:
<dbooth> Meeting: Weekly Baking Club Meeting


WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Got date from IRC log name: 07 Apr 2011
Guessing minutes URL: http://www.w3.org/2011/04/07-hcls2-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