<michel> Fred gave his use case

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

<iker> http://code.google.com/p/lindenb/source/browse/trunk/src/xsl/dbsnp2rdf.xsl

<iker> here is an example of the transformation using this XSLT http://code.google.com/p/lindenb/wiki/Xsltstream

scribenick BobP

Scott discussing possible massive join.

Scott: Trick is to know which xml query to need

Michel: NCBI is Oracle
... We need just the SNPs, do not need to extend further

Michel and Iker discussing xml schema

Iker: XML schema is really long.

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

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

Above link is for the axioms!

Sorry - is for the paper.

Scott: Agree w need to come up with data of some sort
... problem: warfarin is used in many situations; Ex heart valve etc
... warfarin used for many things. So cohort is more general than a given disease
... 'bedridden' is one of the indications, not really a disease cohort

Michel: pharmGKB is from trials data

<mscottm> Above: I was talking about a potentially massive SQL query that would be used to access the key data elements in the SNPdb database (local copy).

Scott: Have large list of to-do. Been contacted by editor re progress.
... deadline is July 7.
... due in three weeks. What are the main bottlenecks?

<iker> +1 timeline

Elgar: Found couple of papers on methodology for predictors for warfarin dosing
... most of papers are in google doc
... tested predictors against other methods out there
... 5 predictors: 2 clinical, 2 gene tables, 1 other
... seems that pgx algorithms best > 50%
... for methods, might be interesting to contact the authors to get data

<epichler> Finkelman et al. 2011

<epichler> Journal of the American College of Cardiology Vol. 57, No. 5, 2011

<epichler> Tatonetti et al. BMC Bioinformatics 2010, 11(Suppl 9):S9

Elgar will contact authors

Scott: Can validate our modelling w their methodology

Elgar: Here are their methods based on these feature vectors

Scott: So using best predictor, then use table method

Elgar: We enable being able to ask certain queries.

<mscottm> Query the knowledgebase: What would be the best predictor? Compare answer to the answer given by the article's recommended method or algorithm

<iker> +1 this approach

Fred discussing integration of data from different contexts and from different purposes

Fred: Considering whether clinically useful or validated

Scott: Talking to clinical genetics dept, to set up decision pipeline
... decisions based on alleles or snps, use linked data for interoperability

<iker> my regrets but i must leave

Scott: extract clinical data, ability relies on the ability to represent the SNPs etc
... even hypotheses might be captured
... seems not practical now to look back 10 years b/c not represented in data
... advantage in modeling is to implement algorithms in a lucid fashion
... if algorithms written using concepts that are traceable, this will be powerful
... clinical data pipeline: one long information flow requires these models and linked data
... model for SNPs can be expanded, here's a way to use the data
... little fuzzy on where comparison is made between querying in linked data, and in table format
... difficult to present to pgx crowd maybe

Fred: Concrete example of who uses and for what would be good here.

We're talking all about data structures :-)

Scott: Need comprehesible explanation for real people.

Fred: Who do you want to read this?

Scott: Propose a way to deal w pgx data and knowledge so can be harnessed in the clinic
... make use of the knowledge in a clear fashion

Fred: Clinician, or researcher, or concerned w some disease?

Scott: Somebody in either form of research; concerned w personalized medicine, lots of people
... general approach to data handling. Futuristic for clinics, just trying to deal w patients
... people in pharma, people in genetics (maybe); pgx is relatively young
... not general practicioners; drug development people, but not drug discovery
... downstream from lead discovery, before you go into staged trials
... Meeting in London NCRI, I heard gene variants everywhere, in all kinds of projects
... gene variants is a bottleneck. w/o info requirement to id, then people are not sure what they need to do
... danger of 100 diff approaches to refer to SNP, fragmentation; another data hurdle
... not evident until you try to share data; but data sharing is inevitable and is the next step

Fred: Sounds like a much needed capability

Scott: Thanks to Fred for sanity check!

This is scribe.perl Revision: 1.136  of Date: 2011/05/12 12:01:43  
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: BobP
Inferring Scribes: BobP

WARNING: No "Topic:" lines found.

Default Present: Bob_Powers, +1.714.616.aaaa, +20416aabb, +1.781.431.aacc
Present: Bob_Powers +1.714.616.aaaa +20416aabb +1.781.431.aacc

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: 16 Jun 2011
Guessing minutes URL: http://www.w3.org/2011/06/16-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