HCLSIG/Pharmacogenomics/Meetings/2013-05-15 Conference Call
Meeting: Clinical Pharmacogenomics Date: May 15, 2013 Time: 10:15 Eastern Time (16:15 Central European Time) Frequency: 1st and 3rd Wednesday of the month Conveners: Michel Dumontier, Matthias Samwald Dial-In #: +1.617.761.6200 (Cambridge, MA) VoIP address: sip:zakim@voip.w3.org Participant Access Code: 4257 ("HCLS") IRC Channel: irc.w3.org port 6665 channel #HCLS
Suggested Agenda
- Simon Lin (Marshfield Clinic) to present/discuss "Big Data Needs in Clinical Genomics". Simon aims to work with the group to collaboratively edit a use case document.
- Continuation of planning and priorities meeting from last teleconference. Prepare for this meeting by jotting down specific outcomes you want to see in the next 3, 6, 12 months (if you have not reported on this already). The minutes of the last teleconference are available at http://www.w3.org/wiki/HCLSIG/Pharmacogenomics/Meetings/2013-05-01_Conference_Call
Meeting Minutes
Topic: Oliver introduces his project of extending BioPAX for representing effects of mutations and pharmaceutical on pathways
[16:22] <matthias_samwald> oliver: i am interested in connecting clinical with molecular data
[16:23] <matthias_samwald> ... also adding kinetic parameters
[16:23] <matthias_samwald> ... how different mutations affect kinetic parameters
[16:23] <matthias_samwald> ... work for PanGenX
[16:23] <matthias_samwald> ... i assume other parties would be willing to consume such data
[16:24] <michel> harmony meeting: http://mbine.org/events/HARMONY_2013
[16:24] <matthias_samwald> ... will be at the harmony (sp?) conference where i will chat with data providers on how the data can be used
[16:25] <matthias_samwald> ... the idea is to extend biopax to represent genetic mutations and their effects on pathways
[16:25] <matthias_samwald> ... also should be able to express that some molecules are drugs given externally
[16:26] <matthias_samwald> ericP: do you also want to represent diseases?
[16:26] <matthias_samwald> oliver: not really at the moment.
[16:27] <matthias_samwald> ... current problem is representing drugs, how they interact with pathways, making exlicit that they are not PART of the pathway.
[16:27] <oliver> zakim, who's here?
[16:28] <matthias_samwald> oliver: for compunds we consider using drugbank, dailymed... mostly drugbank
[16:29] <matthias_samwald> oliver: BioPAX extension will be openly available.
[16:31] <michel> btw - we have sabio-rk in bio2rdf http://sabiork.bio2rdf.org/fct/
[16:31] <matthias_samwald> oliver: idea is to contact data providers to update their BioPAX export
[16:32] <michel> http://download.bio2rdf.org/release/2/sabiork/sabiork.html
[16:32] <matthias_samwald> oliver: signalling gatewaye molecule pages, Reactome are also targets
[16:34] <matthias_samwald> oliver: there is not that much data relevant for clinical pharmacogenomics
[16:34] <matthias_samwald> michel: you know about BRENDA... but it is not openly available
[16:35] <matthias_samwald> michel: do you know of another such database (connection mutations and kinetic parameters)
[16:35] <matthias_samwald> oliver: sabio-rk has kinetic parameters, but does not export data on mutations
[16:36] <matthias_samwald> michel: what about the experimental form?
[16:36] <matthias_samwald> oliver: ???
[16:36] <matthias_samwald> michel: for example a phosphorylated protein.
[16:37] <matthias_samwald> michel: could that be used for mutation representation?
[16:37] <matthias_samwald> oliver: not sure about that. do you have an example?
[16:37] <matthias_samwald> michel: you could simply refer to a database where the variation was described.
[16:40] <matthias_samwald> matthias: sabio-rk RDF is simply the BioPax dump?
[16:40] <matthias_samwald> michel: mainly. did some URI modifications, linking et cetera...
[16:41] <matthias_samwald> matthias: that defeats the purpose of having a BioPAX export a bit.
[16:42] <egombocz> Bio2RDF enhanced SABIO-RK is a nice BioPAX representation for interoperability (example http://bio2rdf.org/page/bio2rdf_dataset:bio2rdf-sabiork-20120918 )
[16:44] <egombocz> As Michel points out, the advantage is interoperability
[16:45] <matthias_samwald> matthias: next steps?
[16:47] <matthias_samwald> michel: i guess you should initiate a discussion with the BioPAX community and the sabio-rk team.
[16:47] <matthias_samwald> oliver: question is how they represent the data internally.
[16:48] <bobP> http://www.w3.org/wiki/HCLSIG/Pharmacogenomics/Meetings/2013-05-01_Conference_Call
[16:49] <matthias_samwald> topic: future plans
[16:49] <matthias_samwald> bobf: i came up with two umbrella topics that we could adress
[16:50] <matthias_samwald> ... ontological CDS -- aggregations of variants of SNPs (star alleles nomenclature etc.), drugs names etc.
[16:51] <matthias_samwald> ... NGS will also become a hot topic.
[16:51] <matthias_samwald> ... need systems to dynamically use rules that can be maintained efficiently
[16:52] <matthias_samwald> ... current systems simply enumerate gene-drug combinations... this is simply not scalable
[16:53] <matthias_samwald> ... reasoners could come up with higher-level conclusions in EHR that could then be used by more simple CDSS system
[16:53] <matthias_samwald> ... Cerner's CBO should be reviewed more thorougly
[16:54] <matthias_samwald> ... not sure how many people are using CBO, review it, then discuss additional approaches for modelling ontologies for genomic CDS
[16:54] <oliver> I need to go to another meeting now. Nice talking to you. Talk to you soon.
[16:54] <matthias_samwald> bye oliver
[16:54] <oliver> bye!
Topic: Continuation of discussion of goals and priorities from last teleconference
[16:54] <matthias_samwald> bobf: need to think about which level (allele, phenotype, snps) are the triggers.
[16:54] <michel> cbo: http://www.cerner.com/about_cerner/clinical_bioinformatics_ontology/
[16:56] <matthias_samwald> matthias: indeed this is a strong point of using OWL and reasoners for this (you can throw any level of abstraction into the reasoner and it can work with it and indentify inconsistencies)
[16:57] <michel> CBO implemented in EHR systems.
[16:57] <michel> bobf: if we could leverage this, we could get it into practice
[16:57] <matthias_samwald> bobf: basing on CBO could speed up adoption.
[16:57] <matthias_samwald> bobf: if we could get cerner on board here, our work could be used in practice quickly
[16:58] <matthias_samwald> bobf: PGRN is taking a very exhaustive approache for manual curation of gene-drug interactions
[16:59] <matthias_samwald> ... but if gene-gene-drug or gene-drug-drug interactions come into play, this approach falls apart very quickly. ontologies could help here as well.
[17:00] <matthias_samwald> ... drug-drug interactions and recommend alternative treatments that are actually better and valid (rather than leaving clinicians to use trial&error to find better treatment)
[17:01] <bobP> (should we just google Cerner CBO?)
[17:02] <matthias_samwald> matthias: mapping to CBO is our best bet, trying to extend/ontologize it seems like a lost cause
[17:02] <matthias_samwald> bobf: good point, maybe we should point out to them what it is needed.
[17:03] <matthias_samwald> ... CBO currently is more of a taxonomy
[17:04] <matthias_samwald> bobf: the Medicine Safety Code (MSC) is a very neat idea, i think we can create some very quick early adoption
[17:05] <matthias_samwald> bobf: PGRN-seq will be used by PGRN and others
[17:05] <matthias_samwald> bobf: need a way of exchaning information across clinical settings. chip is only limited to 84 genes, maybe only a couple of dozen are pharmacologically actionable
[17:07] <matthias_samwald> bobf: MSC demo could quickly be adopted by PGRN and eMERGE networks
[17:09] <matthias_samwald> bobf: should have MSC in place when PGRN-seq goes live
[17:10] <matthias_samwald> matthias: where can we get the PGRN-seq list?
[17:10] <matthias_samwald> bobf: will post into the chat.
[17:11] <BobF> PGRN Seq project update:
[17:11] <BobF> PGRN Seq gene list: