HCLSIG/PharmaOntology/Meetings/2011-07-14 Conference Call/Minutes
* Now talking on #hcls2 * Topic for #hcls2 is: #hcls2 pharmacogenomics paper * Topic for #hcls2 set by michel at Thu Jun 30 11:59:50 2011 * mscottm (marshall@81.69.36.208) has joined #hcls2 * Zakim (rrs-bridgg@128.30.52.169) has joined #hcls2 <mscottm> Zakim, who is here? <Zakim> sorry, mscottm, I don't know what conference this is <Zakim> On IRC I see mscottm, BobP, ericP <mscottm> Zakim, this is tmo <Zakim> ok, mscottm; that matches SW_HCLS(TMO)11:00AM <BobP> scribenick BobP <Zakim> +??P29 * iker (5530ca0d@207.192.75.252) has joined #hcls2 * epichler (ad4c1522@207.192.75.252) has joined #HCLS2 <Zakim> + +1.781.431.aaaa <BobP> Scott: On the way to making a clinical query work <BobP> ... want to do patient data from pharmGKB, by Monday? <BobP> ... Adrien replied <BobP> ... Adrien has warfarin text-mined relations in RDF <BobP> ... pharmGKB could go quickly * michel (c13ece97@78.129.202.38) has joined #hcls2 <Zakim> +??P10 <michel> zakim, ??P10 is michel <Zakim> +michel; got it <BobP> ... (talking re RPI hackathon) <BobP> Michel: Can take over pharmGKB mapping <BobP> ... add column headers from pharmGKB, add to TMO <BobP> ... will add script to TMO svn <iker> two interesting data sources from PharmGKB <iker> https://www.pharmgkb.org/do/serve?objCls=PhenotypeDataset&objId=PA135603152#tabview=tab0 <BobP> Scott: pdfs can go into dropBox <iker> https://www.pharmgkb.org/do/serve?objId=PA161845844&objCls=PhenotypeDataset#tabview=tab0 <BobP> Elgar: Can set up dropbox <BobP> that was fast too! <BobP> Michel demonstrating lightning fast facility w Google * michel - lol <BobP> Scott: Admiring Iker's query <BobP> ... walk-through w competency questions <ericP> Zakim, please dial ericP-office <Zakim> ok, ericP; the call is being made <Zakim> +EricP <BobP> ... clinical is covered. Remains: drug development pharma mining example <BobP> ... ideally we describe how solution links out <BobP> Scott: Center of paper has shifted, but we can describe a pharma ex <BobP> ... vicinity of gene or related genes; easy to move from one aspect to another <BobP> ... on the shelf for now. <michel> zakim, who is making noise <Zakim> I don't understand 'who is making noise', michel <BobP> ericP's mike is not noisy <BobP> ericP: Looking at Iker's query, w replies <BobP> Iker: Not making use of SWObjects <BobP> ... just Xferred dbSNP to RDF <BobP> ... can KB for this be added to TMO KB? <BobP> ... w TMO KB, I can upload <BobP> Scott: Can find homes for dbSNP and for warfarin data. <BobP> ... should be able to refer to KB separately as well as together <BobP> Iker: Should extend TMO, or use another ont, for mappings? <BobP> Scott: Mapping to semantics can happen w SWObjects? <BobP> ericP: Here's how I can construct IRIs in proper form. Iker, do you want to materialize? <BobP> ... URL construction can be done in several ways <BobP> ... if sparql technique doesn't work, then we need to think <BobP> Iker: (discussing tech w ericP) <BobP> ... datamined one line at a time(?) <BobP> ... used D2R2. Issue is, chromosome maps to local vocabulary <BobP> ericP: Issue is not tech, but which vocabs to settle on. <BobP> Iker: Found a vocab for SNPs <BobP> Scott: For choosing terms: gene variant-specific terms would come from our own TMO <BobP> ... once stabilized, then ericP does mappings <BobP> ... long term, most sustainable approach is to create mapping to SWObjects <BobP> ... rather than materialize <BobP> ericP: Change in schema, we would have to adjust anyways <BobP> ... tools already exist <BobP> ... prefer to use my tools, more expressive now <BobP> Iker: D2R fully materialized it <BobP> Scott: As long as we know it;s working, we can write about it. <BobP> ... but we should move more toward sustainable solutions <BobP> ... if NIH has sparql endpoints, then they can move toward happiness w us <BobP> ericP: Can use sqlServer itself <BobP> (microsoft taking hits here :-) <BobP> Scott: ericP you have got the most agile solutions <BobP> ... already have materialized RDF for sparql query <BobP> ericP: Next step, figure out schemas for SNPs. <Zakim> -EricP <BobP> Michel: Adding more terms from TMO, some from SO. Has terminology <ericP> Zakim, please dial ericP-office <Zakim> ok, ericP; the call is being made <Zakim> +EricP <BobP> Michel: W. list of types, then can add to TMO <BobP> ericP: TMO or SO? <BobP> Michel: SO is substantially larger. Either case necessitates mapping. <BobP> ... ontological commitment. Real entity or info entity? <BobP> ... SO commits you to info content, w TMO we can use realist ont <BobP> ericP: SO lots of fringe? <BobP> Michel: Can use mireot to extract minimal ontology, just hierarchy. <BobP> ... real problem is that chromosome is an information content entity <BobP> ericP: Matters in some universe? Proteins as aggregate vs specific prot in a cell <BobP> Michel: There are experiments w individual molecules! <BobP> ericP: Just spell it differently? <BobP> Michel discussing property sillinesss issues down the info path <BobP> ericP: Happy to stick w informational solution <BobP> Michel: Statements about the nature of genes, doen't make sense re info <BobP> Elgar asking about the choice <mscottm> good answer! <BobP> ... inclined to do it this way too <BobP> Michel: One KB material, other KB info, then disjoint and we have to deal with this eventually anyway <BobP> ... if pick SO to do mapping, that's OK for now. <BobP> Scott: Readers of this paper, assuming pgx apps, won't notice the difference <BobP> ... no difference at the query level <BobP> ericP: asking about classes vs individuals <BobP> Michel: Semantics communicates to the user the kinds of things they're dealing w in their KB. <BobP> ... recall that we started w modelling question <BobP> ericP: +1 <BobP> Scott: Community curated ontology in your email? <BobP> ... don't have to decide on SO yet? <BobP> Michel: Will add terms to TMO for now. <BobP> Elgar: Would suggest to go w material entities <BobP> Michel: Safe bet, we have our terminology under our control <BobP> ... we can map everything in dbSNP... <BobP> ericP: -1 Fragmentation of semweb, distinctions should be demonstrated by use cases <BobP> Michel: kebbi has material molecular things, but we have disjoint things * mscottm agrees with ericP on the principle of reuse and the need for a use case that demonstrates the need for a different model <BobP> ericP: Say, new use case exceeds expressivity of model <BobP> ... sort of minimalist, but decisions have already been made. * michel the reuse is satisfied by mapping <BobP> Elgar: We have already followed this paradigm in TMO <BobP> ... we chose to use TMO protein class <BobP> ... would have preferred to use mappings to other onts <BobP> ... might have pulled in SO terms <BobP> ericP: Agree, consistent. <BobP> Michel: Updating TMO in svn, not webProtege <BobP> Elgar: webProtege seems to be a one-time dump <BobP> Michel: Can put up owl-doc version, or ontoB(?) <michel> another ontology viewer: http://www.ontobee.org/ <BobP> Michel: Need list of types to use <Zakim> -EricP <mscottm> Zakim, who is here <Zakim> mscottm, you need to end that query with '?' <mscottm> Zakim, who is here? <Zakim> On the phone I see Bob_Powers, mscottm, ??P29, +1.781.431.aaaa, michel <Zakim> On IRC I see michel, epichler, iker, Zakim, mscottm, BobP, ericP <mscottm> Can anybody hear me? <Zakim> -michel <mscottm> thanks all! bye! <Zakim> -mscottm <epichler> dropped voice connection - see you Sun - e2 <Zakim> - +1.781.431.aaaa * epichler (ad4c1522@207.192.75.252) has left #HCLS2 <BobP> rrsagent, draft minutes <BobP> zakim, who is here? <Zakim> On the phone I see Bob_Powers, ??P29 <Zakim> On IRC I see michel, iker, Zakim, mscottm, BobP, ericP * iker has quit (Quit: http://www.mibbit.com ajax IRC Client) <BobP> rrsagent, draft minutes <mscottm> Oops - I'm afraid that I forgot to invite RRSAgent. <Zakim> -Bob_Powers <BobP> oh * michel has quit (Quit: http://www.mibbit.com ajax IRC Client) <BobP> i can email the minutes <mscottm> Thanks!