HCLS

<matthias_samwald> ack ??P3

<ericP> meeting: HCLS F2F

<ericP> scribenic: ericP

<ericP> leina: U Texas

<ericP> ... pushing experimental data into SemWeb

<ericP> Don Doherty: neurscientist

<ericP> ... started company "BrainStage"

<ericP> Kei: faculty at Yale Center for BioInformatics

<ericP> ... working on SemWeb for integrating Neuroscience info

<ericP> ... member of SenseLab

<ericP> Alan Hobbs: neuroscientist at Kaieser Permanente

<ericP> ... collegue Peter Hendler will be here tomorrow to present SNOMED/OWL

<ericP> Masao: am Keio site manager

<ericP> ... today mostly working on home networks (not connected to SemWeb), but need to bring HCLS message back to Japan

<ericP> Harold Solbrieg: description logics with NCBO

<ericP> ... creating services for access to terminilogical content

<ericP> Micheal Miller: involved with Microarray gene expression standards

<ericP> ... importing data into rosetta net

<ericP> Fahim: neuoroscience information framework at UC San Diego

<ericP> Johsua Phillips: working on CaBig

<ericP> ... exposing linked data to CaBig

<ericP> ... using ontologies to connect data

<ericP> Susie: cochair of HCLS group

<ericP> ... working on LODD and TMO

<ericP> ... do participate on others, like SciDisc

<ericP> ... am director of science and tech at J&J

<mscottm> remote participants, please use the LiveMeeting link for *Monday* at the bottom of http://esw.w3.org/topic/HCLSIG/Meetings/2009-11-02_F2F

<ericP> matthias_samwald: started out in neuroscience, specialized in semweb and ontologies

<ericP> ... working at DERI Gallway

<ericP> ... involved in lots of SemWeb cognitive projects

<ericP> ... creating standards and conventions which can be adopted on a wide scale

<ericP> Nigam Shah: Stanford U NCBO

<ericP> Jeff: San Diego ...

<ericP> Micheal: @@@

<ericP> ... Roche pharmaceuticals

<matthias_samwald> is there a link for the presentation?

<matthias_samwald> http://esw.w3.org/topic/HCLSIG/Meetings/2009-11-02_F2F?action=AttachFile&do=get&target=DonaldDohertyHCLS2009.pdf

<matthias_samwald> see it now

Overview of neuroscience and informatics requirements [Don Doherty]

<ericP> Don: spend a year busy with buisiness stuff

<ericP> ... you got a lot from Maryanne Martone (running NIF) last year

<ericP> ... informatics underrepresented at recent neuro conference

<ericP> [slide 2] [picture of brain]

<ericP> ... most complex piece of matter, examined for 2K years

<ericP> [slide 3] Nicolaus Steno

<ericP> ... "to learn a machine, either the master who made it must show us it's artifice, or we have to take it apart and examine it"

<ericP> [slide 4]

<mscottm> Don: .. once Watson and Crick figured out the structure of DNA that lead to a lot of new developments

<mscottm> Don: .. Linus Pauling figured out the structural chemistry of Vita C

<ericP> Don: haven't had the "double-helix revelation" for the brain

<ericP> ... we can do microarrays in the brain

<ericP> [slide 5] Frueud's project

<ericP> Don: Freud started as an anatomist

<ericP> ... scientists needed independent support, so became a doctor

<ericP> ... wrote up a book on cognitive psychology

<ericP> ... gave up and moved to other means

<ericP> [slide 6] The Project for Scientific Psychology

<ericP> ... LTP ideas from circa 1895

<ericP> ... much of our current research still rolling arouund ideas from centuries ago

<ericP> [slide 7] Santiago Ramon

<ericP> Don: Santiago: information flows from axons to dendrites to axons to dendrites

<ericP> ... mostly true

<ericP> [slide 8] [labeled brain]

<ericP> [slide 9] [sections]

<ericP> Don: starts showing differences in patters of cells in areas of the brain

<ericP> ... we're looking at a very gross level

<ericP> [slide 10] [pointelist brain]

<ericP> Don: in neuroscience, your either a lumper or a splitter

<ericP> ... how do we get concensus and get into the details of how the system works?

<mscottm> [Don] pointing to top of slide 10

<ericP> ... "cortex" means "bark"

<ericP> slide 12] [11, with labels]

<ericP> [slide 12 (guess last was 11a)] gene networks with brain networks

<ericP> don: exampined long range connections in 40s

<ericP> ... (gross connections between brain structures)

<ericP> ... 80% have proven incorrect

<ericP> ... keeps growing

<ericP> [slide 13] Databases and Datasets

<ericP> Don: lots of databases out there now

<ericP> ... seeing useful new techniques, e.g. brainbow

<ericP> [slide 14] New Techniques to See Wiring

<ericP> Don: allows you to see connections in the cells

<ericP> ... gathering data with automated photon microscopes

<ericP> [slide 15] Connectome

<ericP> Don: huge amout of innovation to connect the automated collection with our informatics

<ericP> ... connectome examinines signals on a long range connections

<ericP> [slide 16] Neurophil

<ericP> [slide 17] [purple dye showing intra-cell connections]

<ericP> [slide 18] set of individually traced connections

<ericP> Don: there's growing automation

<ericP> ... proteins in and structure of axons and dendrites all relevent

<ericP> ... processing therein controls how the nerve cell gets info from a to b

<ericP> [slide 19] [more traced connections]

<ericP> [slide 20] [still more traced connections]

<ericP> don: complexity in these microstructure is immence

<ericP> ... illustrations in slide 20 are just cartoons [gross simplifications]

<ericP> [slide 22] Large-Scale Neural Circuit

<ericP> Don: manual tracing would require 10Ks of man-years

<ericP> [slide 23] [microscopy image]

<ericP> Don: this image is on a 20 micron scale (of 3lbs of brain in a human)

<ericP> [slide 24] [Signals in the Neurophil]

<ericP> [slide 25] [Signals in the Neurophil]

<ericP> [slide 26] Activity in Awake Behaving Animals

<ericP> [slide 27] Neuroinformatics Challenge

<ericP> Don: is old new, a montra

<ericP> [slide 28-29] Some Exciting Tech Projects

<ericP> [slide 30] Brain Navigator

<LenaDeus> Brain Navigator: http://www.brainnav.com/

<ericP> Don: saw Elsevier offering Brain Navigator at conf@@3

<ericP> [slide 31] The NeuroNetwork

<LenaDeus> Neuronetwork: theneuronetwork.com

<ericP> Don: Springer offering a Facebook for neuroscientists

<ericP> ... expect useful tool will lie in the middle

<ericP> [slide 32] Summary

<ericP> Don: have so much, how do we apply to a broader audience

<ericP> Micheal, end in site to data growth?

<ericP> Don: exponential growth over 50 years

<ericP> ... Connectome should give us a better sense of "miso-scale" connections

<ericP> ... at the micro-scale, we're at the very beginning

<ericP> ... as with science, we thought that DNA was the end, but now we have RNAa...

<ericP> Kei: thanks for the talk, touch on size and complexity

<ericP> ... are you taking variation into account?

<ericP> ... is this a canonical brain?

<ericP> Harold: should be Ventner's brain

<ericP> Don: Connectome starts with a mouse

<solbrig> Ventner

<nigam> meso-scale --> meso-scale

<nigam> Cool! .. i can do regexp

<ericP> Don: a brain structure within the brain is amazingly specific

<ericP> ... stereotypes work; we can find cells in one worm and find corresponding cells in ever other worm

<ericP> ... can use new techniques to get down the the details of e.g. fruit flies and worms

<ericP> ... i think it makes sense to focus on a fly or a worm (though many folks not interested in a work)

<ericP> Alan Hobbs: will a @@5 help us extract variations?

<ericP> ... and help clinicians, etc.

<ericP> Don: e.g. ANDI looks at neurodegeneration over time in alzheimer's patients

<ericP> ... seen how individuals, or [characteristic] groups are important

<ericP> ... don't see many pubs which differentiate a male and female brain

<ericP> ... despite obvious dimorphisms; cells that exist in one or another gender

<ericP> [missed michael miller's comment]

<ericP> Don: usually discussion of sex differences concentrate early hormonal differences

<ericP> ... how able to move genes off of X and Y to differentiate between hormones and physiology

<ericP> mscottm: great article illustrating examining a physical components

<ericP> [slide -- picture of a PC board in a radio]

<LenaDeus> can a biologist fix a radio: http://protein.bio.msu.su/biokhimiya/contents/v69/pdf/bcm_1403.pdf

<ericP> [slide -- picture of a two PC boards in a potentially two radios]

<ericP> [slide -- simple electronics schematic of a radio]

<ericP> mscottm: i don't think it's hopeless for us to discover how brain works by examining components

<ericP> ... we're not looking at the transitors to study transmitted music and culture

<ericP> ... instead we're grouping them to L-C circuit, etc.

<ericP> ... need to share that data across disciplines

<ericP> ... we'd like a SemWeb desc of the components (||- resistors, capacitors, ...

<ericP> ... ).

<ericP> ... want to predict what happens if we change one of these components

<ericP> ... leads to a question:

<ericP> ... .. you said we want to gather data and integrate it

<ericP> ... .. what are the bottlenecks in that process?

<ericP> ... .. are we waiting on anatomical ontologies?

<ericP> ... .. are people at different levels of focus but need to align?

<ericP> Don: so many areas, level of detail varies

<ericP> ... any framework we offer may be proven incorrect

<ericP> ... as informaticians, we need to be flexible to changes in understanding

<ericP> ... at Brainstage, we've worked hard on usability

<ericP> ... is it both usable and fun?

<ericP> ... places like facebook entice folks offer content

<LenaDeus> eric:is there an infrastrucure we can develop to improve people collaborating

<LenaDeus> eric: if we recordd a piece of data on the SW and then new discoveries come along, that is where the SW will prove itself - will data still be useful?

<LenaDeus> eric: should we prove that to the world (data still be useful even though it was incorrect)?

<ericP> mscottm: should we characterize different types of research?

<ericP> ... you have simulation folks, who could try to express e.g. differential equations in RDF

<LenaDeus> BlueBrain: http://bluebrain.epfl.ch/

<ericP> ... when speaking with the BlueBrain folks, you want to speak about components

<ericP> ... you can combine semantic annotations on these components with acquired data [e.g. bulk relational data?]

<ericP> Don: many folks beleive that ultimately we will need mathmatical models to understand the brain

<ericP> ... there was a recent paper mathmatically describing neuron morphology

<ericP> mscottm: trying to see what we should offer; do we need to cater to e.g. modelers?

<ericP> ... there's also the omics viewpoint

<ericP> nigum: are we trying to sell SemWeb tech to solve existing Neuroscience problem? or offering a new system?

<ericP> ... service mindset projects are good fodder for companies

<ericP> ... not as good for impressing folks

<ericP> ... wow factor: we can do something you can never do before

<ericP> mscottm: need to be pragmatic. if a service-oriented opportunity presents, we should consider it

<ericP> ... in 92 i asked "why is there no such thing as bioinformatics?"

<ericP> ... (now terabytes of data)

<ericP> ... a lot of projects will produce lots of data, but there's no point if the data isn't shared

<ericP> ... need to provide for data stewardship

<ericP> ... even in systems biology, proposals offer [clunky] interfaces to e.g. RDBs

<ericP> ... just making data accessible, it's more than they've had

<ericP> Michael: was impressed by SenseLab 'cause it's available today

<ericP> ... SemWeb islands becoming available and connectable

<ericP> ... new RDF tags will help websites become SemWeb-enables

<ericP> mscottm: SemWeb doesn't require same distribution as document web

<ericP> ... SemWeb works well for latent connection of islands

<ericP> ... which islands do we need to connect?

<ericP> Kei: Don showed lots of examples at the end [slides 28-29]

<ericP> ... they're not currently connected

<ericP> ... could allow scientists to ask questions that can't currently

<ericP> ... need to develop ontologies which offer value to neuroscientists

<ericP> Mark Musen: parallel between work on understanding brain function and our informatits work

<ericP> ... big space between meso-level and micro-level

<ericP> ... there's no ontology for systems at this level

<LenaDeus> Mark Musen: ontologies are useless in neuroscience because we don't know about all the structures, let alone how to name them

<ericP> ... could encourage research communities to produce their own onts and then fit them into the bigger picture

<ericP> Allen Hobbs: need a vocabulary in which they can express their questions

<ericP> ... need a forum jsut for the SemWeb

<matthias_samwald> (voices from the 'audience' are a bit difficult to understand through Zakim, in part)

<ericP> Don: most of the folks interested in their one area

<ericP> ... the high-profile folks are interested in e.g. Connectime

<ericP> ... everyone now knows about the SemWeb

<ericP> ... Connectome folks talked about OWL and cited SemWeb

<ericP> ... W3 has gotten this process kick-started

<ericP> ... don't need to market as much as help

<ericP> ... NIF is pulling this info together

<ericP> ... the bulk of folks don't go the informatics posters

<ericP> ... scientists just ask "when can we use it?

<ericP> Allen: how do you get beyond "i want to use it now"?

<ericP> Don: there are informatics meetings at the conferences, but same faces

<ericP> Fahim: within NIF, there are knowledge engineering folks and domain folks

<ericP> ... gap between the two

<ericP> kei: i still have a hard time explaining URI to Gordon

<ericP> Fahim: Neurolex.org is a wiki format of our core ontology, NIF standard

<ericP> ... currently export OWL ont into a wiki, but no mechanical return

<ericP> ... we have a cell working group who was able to start with small annotations

<ericP> ... (instead of learning a technology)

<ericP> mscottm: how do you use them? outline views? annotate stuff on wiki? SPARQL-able?

<ericP> Fahim: neurolex more of a wikipedia environment

<ericP> ... have category pages for ontological concepts

<ericP> ... can explore through tree-like views, or export CSV dumps of the graph

<ericP> ... easier to learn than Protege or DL

<ericP> ... developing simple forms, e.g. info on a neuron in a brain region

<ericP> Harold: i can chat for 15 mins about how Neurolex works

<ericP> Michael: i like the "can folks use this?" question

<jun> hello everybody:)

<ericP> ... concentration on getting the terminology

<ericP> Michael: doesn't have a business use

<ericP> Harold: if your familiar with wikis and wiki annotation, it gets you RDF triples of that for very low cost

<ericP> ... it's isomorphic to RDFa

<ericP> ... captures knowledge to the extent you have it

<ericP> nigam: look wikipathways

<ericP> ... wiki used for multiple uses of concept annotation capture

<ericP> ... vs. data annotation capture

<LenaDeus> eric: can I do query federation on top the the wikis?

<ericP> ... wikipathway offers a wiki interface but forces use of controled vocabs

<ericP> michael: when i'm doing an experiment, i'd like to see who else is doing that?

<ericP> nigam: wikipathways (UCSF) is similar to wikiprofessional

<ericP> ... wikipathways offers editing of the ontology itself at the same time

<ericP> karen: how can we offer something to help build this community?

<ericP> ... we should think about our communications roll

<ericP> ... we should be conneting more at neuroscience conferences, e.g. BOFs

<nigam> ... wikipathways allows you to edit a pathway in a visual environment that "forces" you to speak in GenMAPP or BioPAX without being bogged down in the syntax for those pathway ontologies

<ericP> mscottm: there are several wiki's have been mentioned in this room

<nigam> ... then it allows you to annotate the captured pathway using terms from several controlled vocabularies (loosely called "ontologies)

<ericP> ... folks in W3 are looking at new semantic wikis, but would be nice to re-use a common framework

<ericP> harold: i think all i've heard of are, or will, be mediawiki

<nigam> ... the paradigm is the same as wikiproffessional (that of using the wiki-platform for collaborative knowledge capture; or for collaborative annotation)

<ericP> mark musen: everyone knows how simple it is to use wikipedia

<ericP> ... belies the concept that there are usefully modeled data underneath

<ericP> ... think we need to focus on making complex information viewable/editable

<ericP> harold: a lot of the success of the genome project came from the extensibility of the ASN-1 model

<nigam> ... and have clarity on what is the complex information we are making viewable/editable (the ontologies itself, the data, annotations on those data, the discourse about the conclusions derived from the data and so on)

<nigam> ... not making the distinction leads to what I call "wikimania"

<mscottm> Hi Matthias, Jun, Susie - taking a 15 min break

<matthias_samwald> okay, thanks

<jun> thanks

<matthias_samwald> I'l drop off the call til then.

<jun> hi scott

<jun> it was Jun

<andreasplendiani> Hi, I'm having my mic on mute, but I'm listening

<LenaDeus> we're at break

<andreasplendiani> I hear :)

<jun> you need safari for Mac, I think

<andreasplendiani> Yes, it works with Safari. Had problem with firefox

<ericP> Mark's talk

Collaborating with the National Center for Biomedical Ontology

<andreasplendiani> ok, works

<mscottm> Matthias - I'm downloading the pptx and will convert it for you. Shall I put it in .ppt?

<ericP> matthias_samwald, ETA on PDFL 5 mins?

<ericP> 2 mins?

<mscottm> shall I make pdf instead?

<ericP> Mark: invite interruptions inline

<ericP> [slide 2] NIH Roadmap

<matthias_samwald> pdf is okay

<ericP> [slide 3] NIH Roadmap (geographical)

<ericP> [slide 4] NCBO

<ericP> Mark: make ontologies useful, share tools

<ericP> [slide 5] NCBO: Key activities

<ericP> [slide 6] [NCBO Website bioontologies.org]

<nigam> www.bioontology.org

<ericP> Mark: lots of tools for collaboration

<ericP> ... creation of infrastructure, and proseletizing use

<ericP> [slide 7] [bioportal.bioontology.org]

<ericP> Mark: all of the important ontologies

<ericP> ... ... in biomedicine

<ericP> ... can see activity:

<ericP> ... .. recent uploads

<ericP> ... .. annotations

<ericP> ... .. mappings

<ericP> ... developmental ontology has a zillion new mappings

<ericP> ... helps folks where action is

<ericP> ... can subscribe to RSS feed

<mscottm> Mark's talk now is available as pdf (main link).

<ericP> [slide 8] BioPortal allows us ...

<matthias_samwald> thanks

<mscottm> http://esw.w3.org/topic/HCLSIG/Meetings/2009-11-02_F2F?action=AttachFile&do=get&target=MarkMusenHCLS2009.pdf

<ericP> Mark: online mechanism for peer review

<ericP> ... make peer review democratic

<ericP> [slide 9] [bioportal library]

<ericP> Mark: alphabetical list of ontologies

<ericP> ... we are ecumenical about supported onts

<ericP> ... we allow users to filter based on group (e.g. OBO foundry, WHO)

<ericP> [slide 10] [snippit of NCI thesaurus]

<ericP> Mark: fancy animation on right, standard tree browser on left

<LenaDeus> http://bioportal.bioontology.org/

<ericP> [slide 11] Users can view and create mappings

<ericP> Mark: expect to have onts well beyond UMLS

<nigam> http://bioportal.bioontology.org/visconcepts/40644/?id=Skin_Neoplasm#t_tab1:

<ericP> ... have tools which lexically map ontologies

<ericP> ericP: you've spot-checked this for sanity?

<ericP> Mark: yes, sacrified recall for accuracy

<ericP> ... a lot of mappers don't perform

<ericP> Michael: a lot of open source mappers work at toy level

<ericP> mscottm: i think you meed demos

<ericP> Mark: mapping tools which are described at conferences

<ericP> [slide 12] Biomedical Resource Ontology

<ericP> Mark: ontology of algorithms, ontologies, ...

<ericP> ... like a google sitemap but for biomedical community

<ericP> ... BRO developed with collab between NCBO, NIH

<ericP> [slide 13] "Note" in BioPortal

<ericP> Mark: david went through much of the BRO and left "marginal notes"

<ericP> ... have threaded discussions where david and peter argue

<ericP> mscottm: do you use Information Artifact Ontology

<ericP> Trish: can out of a branch of OBI

<ericP> Mark: need to harmonize with that and with NIF

<ericP> ... as with folks who use wikis to comment on ontologies, we use notes, which have more structure

<ericP> ... goal is that ontology developers can look at the discussions and improve the ontology

<ericP> [slide 14] Community-Based Annoataion as Peer Review

<ericP> Mark: expect that users are in the best position to evaluate ontologies

<LenaDeus> eric: Users would contribute comments to improve ontologies because future versions might cover their needs more closely

<ericP> Mark: expect to put TMO up and see if folks leave notes

<ericP> [slide 15] BioPortal is building...

<ericP> [Mark sneezes]

<ericP> Mark: helpful to know when new versions will be released

<ericP> ... use not only for browsing but for finding services

<ericP> ... theis functionality is implemented as web services

<ericP> ... (not just [browser-oriented] web interfaces)

<ericP> [slide 16] Word Add-In

<LenaDeus> http://ucsdbiolit.codeplex.com/

<ericP> [slide 17] Word automated recognition of ont...

<ericP> Mark: it uses Bioportal web services (news to us) to do term matching

<ericP> [slide 18] BioLit Web resource...

<jun> what happens if the terms were changed or deleted?

<ericP> Susie: like the word plugin, would like it for excel

<ericP> mscottm: note Anzo (Cambridge Semantics)

<ericP> Susie: need it to be part of a larger tool set

<ericP> Mark: this area has gotten much more crowded since we got our NIH grant

<ericP> [slide 19] IO informatics

<ericP> Mark: give access to heterogeneous data sources, including stuff indexed in bioportal

<ericP> [slide 21] Ontology Tags

<ericP> Mark: makes sure folks are using appropriate terms on the wiki

<ericP> [slide 22] ISAcreator

<ericP> Mark: comes from EBI

<ericP> ... downloads onts from BioPortal and EBI

<ericP> Michael: mostly for omics and multi-omics

<ericP> ... we're working on a similar tool

<nigam> ... ISA_creator uses the search service as well

<ericP> [slide 23] ECG Gadget

<nigam> All services accessed by these tools are at: http://www.bioontology.org/wiki/index.php/NCBO_REST_services

<ericP> Mark: take a digitized ECG and annotate it with onts found at bioportal

<ericP> [slide 24] Biositemaps editor

<ericP> Mark: when folks enter info on a resource, they try to use controlled terms

<ericP> ... looks at the relevent ontology and integrates terms from the BRO directly into the webpage

<ericP> ... controled web page input theme shows up in this site and Gadget

<ericP> [slide 25] Protégé BioPortal...

<ericP> [slide 26] Simbios

<nigam> The general theme is that we provide REST services to access, download, traverse, search, visualize and refer to ontologies and ontology terms

<ericP> Mark: used to model everything

<ericP> ... our BioPortal widgets give you tools to autocomplete

<ericP> ... having a hard time seeing who's using it

<ericP> ... we don't want to be in the customer support business, or be intrusive, but we need to justify our funding to NIH

<ericP> ... have seem annotations applied to acquired data

<ericP> ... looking at automating annotating experimental data

<ericP> [slide 27] NCBO Annotator

<ericP> [slide 27 build 2] NCBO Annotator...

<ericP> Mark: want to take embedded natural language and do term resolution/disambiguation

<ericP> [slide 28] NCBO Annotator

<ericP> Mark: want to analyze text and get a term cloud

<ericP> [slide 29] Resource Index

<nigam> more on the annotator at: http://www.biomedcentral.com/1471-2105/10/S9/S14/abstract

<ericP> Mark: allows us to search online for related datasets or do datamining from data sets

<nigam> more on the resource index at: http://www.biomedcentral.com/1471-2105/10/S2/S1

<ericP> [slide 30] Example: Indexing GEO

<ericP> [note builds]

<ericP> [slide 31] [NCI thesaurus again]

<ericP> Mark: can match terms for melanoma, but not "skin cancer"

<ericP> ... brings you directly from ontologies to data

<ericP> kei: got a list of gene lists ...

<ericP> nigam: in previous HCSL presentation i showed xml

<ericP> Mark: annotation has become a big bisiness

<nigam> Talk with XML outputs of services: http://bioportal.bioontology.org/visualize/40644

<ericP> [slide 32] GMiner

<nigam> oops .. this one: http://www.stanford.edu/~nigam/cgi-bin/dokuwiki/doku.php?id=download

<ericP> [slide 35] See our wiki

<mscottm> http://www.bioontology.org/wiki/index.php/Using_NCBO_Technology_In_Your_Project

<ericP> http://www.bioontology.org/wiki/index.php/Using_NCBO_Technology_In_Your_Project

<ericP> [slide 36] way to collaborate

<ericP> Mark: you can be a driving bilogical project

<ericP> ... use our grant to offer challenges

<ericP> ... need sweet point between one DBPs and too many DBPs

<ericP> ... can also write NIH research grants

<ericP> ... have had some industrial grants

<ericP> [slide 37] Current DBPs

<ericP> Mark: had to pick 3 of 9 excellent candidates

<nigam> Link to "invasive skin neoplasm" concept to see the visualization tools http://bioportal.bioontology.org/visconcepts/40644/?id=Invasive_Melanoma#t_tab1:

<ericP> Mark: has created a Sesame Street for ontolgies

<ericP> [slide 38] Current "collaborating R01" grants

<ericP> Mark: funding rate is around 80%

<ericP> ... $250-$500K annually

<ericP> ... 3-5 years

<ericP> ... usual annual report

<ericP> ... stimulus funding requires quarterly

<ericP> mscottm: contacts with cleaveland clinic?

<ericP> ... Mark Wilkinson (UBC)?

<ericP> Mark: don't know if the cardio folks are reaching out to cleveland clinic or UBC

<ericP> ... don't need to be an academic to collaborate

<ericP> ... nextbio, google, colbix (SP?)

<ericP> ... google funding web-o-trust

<ericP> ... ontology metadata v @@6

<mscottm> http://ontoware.org/projects/omv/

<ericP> [slide 40] other NCBO components

<ericP> [slide 41] NCBO needs HCLS

<mscottm> http://www.google.com/url?sa=t&source=web&ct=res&cd=2&ved=0CA4QFjAB&url=http%3A%2F%2Fwww.neon-project.org%2Fweb-content%2Fimages%2FPublications%2Fneon_2009_d832.pdf&ei=GjfvSrfzJ4qOswO3ofHhBw&usg=AFQjCNEr5tXE7PoBDFJcw08-UKoQ_ElSWw&sig2=wPF_fux73UzoIAlq2DqYGQ

<ericP> Mark: need you guys to use our tech and drive our requirements

<mscottm> http://neon-toolkit.org/wiki/Main_Page

<ericP> ... use HCLS for proseletizing

<nigam> ... come to the ISMB Bio-ontology SIG to show your work!

<ericP> ... frustrated that W3C recommendations aren't having the traction we'd like

<ericP> lena: if you have an affymetrix micro array, we can plug in a matlab a compontent to get ups and downs automatically

<ericP> ... any term set for such workflow detail?

<ericP> nigam: cytobank is using terms from maybe 25 terms

<ericP> michael: we didn't do it formally; we used apt tools or matlab, but we never thought to hook them up to terms

<ericP> nigam: if someone uses taverna, you can terms from us, algorithms from someone else

<ericP> Mark: had conversation with Carol Goble about this

<ericP> lena: rather than pushing data to the analysts, why not push the tools to the biologists?

<ericP> mscottm: missing experimental conditions as ontologies

<ericP> LenaDeus: protocols which the machine controls are pretty standard

<ericP> mscottm: one of the biggest factors of microarray data is the folks who did the clustering

<ericP> ... most people are stuck just on trying to find corresponding experiements

<ericP> MammaPrint

Discussion: Alignment with NCBO (including ISMB SIG)

<ericP> mscottm, obvious thing is to incorporate your services into our demos

<ericP> nigam: will be offering RDF exports of as many ontologies as we can

<andreasplendiani> RDF export of REST services!

<ericP> ... it will be bio2rdf, but you can tweak the namespace

<ericP> ... how do we plug in our ontology content and reource annotation services [into hcls demos]?

<ericP> ericP: i think you need to attack our use cases 'cause most of what we do is driven by use case

<nigam> RDF exports of ontology-content

<ericP> mscottm: you want to start sharing work at the differentially expressed gene sets

<Karen> +1 Matthias!

<ericP> mscottm: part from gene lists, matthias_samwald uses obo ontologies in his atags work

<ericP> ... would it be useful to access the NCBO services?

<ericP> matthias_samwald: been on the todo list

<ericP> ... was waiting for the OWL version of NCBO services

<ericP> ... the current OWL is very complicated (property restrictions), and the URIs don't match the official obo terms

<ericP> nigam: you're using a prototype

<ericP> ... our production is domain-specific XML 'cause folks have a hard time using the OWL which would require anonymous classes

<andreasplendiani> What about having output in RDF ?

<nigam> Do not use the output of the prototype services!!

<ericP> matthias_samwald: might look good in Protégé, but hard to use

<nigam> All production services are at: http://www.bioontology.org/wiki/index.php/NCBO_REST_services

<ericP> ... happy with a subset of the info, e.g. a list of URIs

<andreasplendiani> Would be nice to have the results in RDF :) (it's not difficult to convert, true...)

<nigam> Need to get full URIs in results

<nigam> Andrea, would you be willing to look at the current XML and propose the RDF you would want

<andreasplendiani> I'll have a look

<ericP> ACTION: nigam and matthias_samwald to work out URIs in OWL format [recorded in http://www.w3.org/2009/11/03-hcls-minutes.html#action01]

<ericP> Susie: could collab on BioOntologies SIG at ISMB

<ericP> ... good opportunity for HCLS to support the SIG

<ericP> nigam: want to steer the SIG towards solving real probs rather than find-grained discussion of terms

<andreasplendiani> bye. Not sure I can login later.

<jun> bye everybody. Have a good lunch

<andreasplendiani> (dinner)

<ericP> hye, susie is here

<egonw> hi ericP, Susie

<LenaDeus> scott: we don't know how huntington disease works, even though we know the gene

<LenaDeus> having lots of data has not helped understand the mechanism

<LenaDeus> data stewardship: doing what we can with the data, not just assembling massive quantities of it!

<ericP> State of the Interest Group: HCLS

<ericP> [slide Knowledge Capture]

<LenaDeus> the way knowledge is extracted from the wet lab is very different from extraction from papers

<matthias_samwald> you have not started yet, did you?

<matthias_samwald> oh sorry, now i see new messages. ooops.

<mm> mm is michael miller

<Karen> scribenick: karen

Scott: Our current task forces
... there are sic

s/six

scribe: BioRDF will be focus tomorrow
... Clinical Observations Interoperability (COI) will give presentation at AMIA
... why they could not be here
... Linking Open Drug Data
... Scientific Discourse is related to soft knowledge we were discussing before
... when people enter a comment into a blog and you want to relate to ontologies and raw data
... Explain the knowledge base
... We wanted to ask a scientific question relevant to data for that question
... We got input from the Alzheimer's Forum
... Drugable targets; signal transduction pathways
... we assembled a number of heterogeneous data sets
... using RDF and OWL
... coupling different sources by choosing same names
... like "genes" and so forth
... used permanent URLs
... the main work was designing that coupling among the different data sources
... The query was fairly complex
... To flesh out restriction classes, to get to sub classes part of the process
... we did pre-inferencing and added entailments to the graph
... so we could ask questions
... Question went across a number of data sets
... and then out came 42 genes
... And AlzForum said it was a great list
... Unlike Google where you came up with 225K hits
... So we have that knowledge base now
... We are building on it
... It is hosted by DERI Galway
... and Adrian Paschke in Berlin
... Also working on query federation
... and there was recently a publication
... in BioInformatics
... where we discussed a combined effort
... looking at federating resources
... applying SemWeb library resources
... will be a great "magic bullet" with a SPARQL query
... different terminology than what's stored in SQL
... storing terminology and talking to relational databases
... Traditional Chinese medicine is part of our triple store federation
... Looking to how that can be used to address problems
... Linking Open Drug Data
... I think you are aware of this one
... differences between document oriented and assertion oriented
... LODD sets
... that have been produced by these activities
... Look at bottom area of the linked data cloud, is all HCLS data
... So it accounts for quite a lot of the LOD data being served up
... I asked Kinsgley Idehan what he considered to be large
... they are serving 7.5 billion triples
... We have moved from millions to billions
... Doesn't mean you can reason over it
... But you can store it and query it
... Ontotext
... at BioIT World talked about what they have
... Ontotext has something like 15 billion before they added tail mix and inferencing
... which moves to 20 bil
... Four Store
... Teratriples

<matthias_samwald> 4store

scribe: Triple stores have gotten quite large
... Uniprot accounts for quite a lot of triples
... Descibed how many triples they process in three weeks
... on order of 3 billion
... Get the idea
... Clinical Operability TF is going strong
... It's work is being shown at AMIA
... It bridges CDSC to HL7 formats
... Eric can talk more about this
... Looking for patients who are eligible for certain clinical trials
... drug companies now have to do foot work
... to look for clinical trial people

Eric: Kaiser Permanente does a good job with this data

Scott: This demo takes data sets and puts in proper formats

Eric: Relational database with hospital data
... HL7 is one view of that
... another view is to ask a CDSC query
... so it pushes queries all the way back to data base
... ontological agility
... without performance issues
... don't have to marterialize it

Scott: This "precipitous gap" between the CDSC expression of elibility and what HL7 can consume
... So take an inclusion criterion
... and translate into a tripe (meformin}
... so we can create a triple to represent that constraint
... SPAQRL query includes that constraint

Eric: There is backing material now
... a Web interface that Helen Chen and Holgar Pichlar put together
... so you can include exclusionary numbers
... you can abstract, not just brand name drug
... It's a realistic usage now

Scott: This is how we perform negation

Eric: I'll talk about SPARQL later today

Q: Is option an outer join?

Eric: yes

Scott: Eric mentioned this drug ontology that's incorporated into the Web interface
... so you can check out the demo live on the server
... In the Terminology Task Force
... links to the BioRDF and federation
... relate features to data
... In various domains, there is a way to relate data to the domains and feature sets
... Text mining and features sets
... gene expression and profiles
... When you have pattern recog algorithms for a domain
... you can classify particular concepts
... good for automated annotation
... People have done things with ontology, Go, and so forth
... In Terminology TF, we decided to employ SKOS
... and not OWL
... Peter can tell us more about that tomorrow
... So we represented SnoMed in SKOS
... There were some Web services that could deal with SKOS out of the box
... AIDA, so we could browse the contents
... John Madden from Duke Univ. showed this at CSHALS conference this year
... Can drag up and have associated labels searched

<ericP> [slide SNOMED CT/SKOS]

Scott: And another view

<ericP> [slide next one afer that]

Scott: Taking it a step further in the federation work
... We were able to directly connect to HCLS knolwedge base at DERI
... then you could browse
... See Go and ? activity
... If you want to create a SPARQL query, which most people do with data bases
... It's hard to write a query if I don't know what's in the knowledge base
... So do knowledge complemtion
... It finds it via label

<masao> i an not open the StateOfHCLS.ppt .. only my problem?

Scott: When you add to query, see it has a PURL URL

<ericP> [slide Access to triples in Taverna]

Scott: Taverna via AIDA plugin
... We want to make it easy to do so it's unavoidable
... to make a semantic workflow
... take a concept and drag it to your workflow
... and say the data coming from here is a gene list
... gene list is a data type defined in OWL, for example
... Just to get the idea, you can drag a label and tag your label
... so access from within the workflow itself
... If something is an integer, which is all you know
... that it's also a chromosome number
... unless you use Biomoby
... This business of pointing at whatever machine is hosting the instance
... HCLS knowledge base is Virtuoso
... Go to these links
... More accomplishments that you can see
... We have gone millions to billions; the LOD cloud is large
... A few months ago a colleague and I gave a course, half of which was SemWeb
... to bio informatics PhD students, so times are changing
... If we can get to point where some of us are using shared names for obvious things, that would be good
... Concept alliance
... The IG itself has six task forces
... Maybe a philosophical question, but a reality
... six TFs is a lot
... Would also like to reach out the Asia
... Europe and America have active participants
... The times of day in Asia have been difficult
... and how many resources to spread around
... It takes overhead to collaborate

Kei: And some people join multiple groups

Scott: Eric, Susie, and I try to spread ourselves across the groups
... and it's great that we can integrate across the TFs
... I hope to see that continue
... I would like to mention this concept that Mark Musen brought across
... If you provide your knowledge and your algorithms as services
... then it allows people to create applications based on those services
... so they get the best verion or release and create agile solutions
... I saw an application that incorporates the BioPortal ontologies in a JAVA app
... and I showed you AIDA
... and we talked about HCLS
... If you have to compile code, it becomes too much overhead to colllaborate
... A SPARQL end point on every talbe
... like a "chicken in every pot"
... If you had a SPARQL end point for Uniprot
... which they would like
... and you have SWAN/SIOC
... and myexperiment
... you could link to workflows
... if you expose RDF
... and these various bioinformatics resources and wikis then you can link together with ontologies
... of course the wiki being more than a wiki
... Not just a page everyone can edit
... Two different things come to mind
... A wiki page could be a view into a knowledge base generated by a template
... and the template edited by people who want to change the view
... The source of info
... if you are able to site a particular resource
... you can create and environment where people site each other and give credit
... this new publishing era that people are discussing

Eric: You may want to mention Workshop

Scott: We just had a Workshop in Wash, DC
... very successful
... fascinating converations
... at ISWC, great presentations
... put it on irc here
... One of central themes was soft facts as an assertion
... Something that's not an assertion in a knowledge base
... but uses literature or something and you want to use that
... so provide knowledge provinence

<ericP> WORKSHOP ON SEMANTIC WEB APPLICATIONS IN SCIENTIFIC DISCOURSE

Scott: Like the SWAN forum, for example
... So if I know that group does good research in Alzheimers, then I may cite that
... This use case for a Semantic Wiki
... Because you can query across multiple wikis
... if they provded these services
... and a user said something about Drug A having a side effect; and side effect similarity with Drug B
... and boost theory
... Drugs with similar side effects have similar targets
... So final slide
... What do we need
... Data stewardship; we need a change of attitude
... The reviewers will hopefully start giving priority to proposals that talk about long-term access
... pay more attention to data colleciton, not just provide an interface
... talk abut how they will do it
... We also need identifiers
... Unclear to me if we have standard identifiers for the most mundane things
... that we can rely on
... so everyone agrees on
... Also scalable triple stores so we can reason in place
... SemWeb isn't magig

s/magic

scribe: Maybe we can be satisfied with lightweight or incomplete reasoning
... accept that I am only 70% finished across a graph
... And we need to coordinate and cooperate across the groups doing all the work
... Would like to come out of this meeting with concrete action items
... Align on the metalevel
... and make sure we are complementing and reusing each other's work
... So that's the state of the Interest Group
... Happy to hear comments
... Coordination and cooperation across groups leads into the next topic
... About outreach
... and how to get more out of less

Howard: not sure how to say this
... What impressed me about this community is the wealth of tools and resources available
... and how isolated they seem to be
... COI group connections
... and caBIG community
... It does boggle me how little I see these resources included in grants and places
... where they can benefit from them the most
... I work with National Cancer Institute
... this may seem too advanced to them
... but what do we do so we can start using all this tuff
... I see you, Mark[looks at Mark Musen]

Nagin: What is good for one may be disastrous for others

Mark: You may be right that some are scared of this
... I think SemWeb community has done good job of outreach

Howard: I would like to see some of work done in grant proposals

Scott: I mentioned some of European partners
... I would include more from States in these projects if I could
... Idea is to provide data to neuroscientists
... by means of good search support
... instead of using traditional techniques only
... you cannot ignore data warehousing; it works well for some situations
... will always be another data warehouse you need to include
... use for the small stuff
... some data is growing rapidly
... leave where it is, federate it

Howard: My perception that it's not being used; that it's isolated?

Mark: There is some isolation
... caBIG has isolated itself
... and OMG is in another world

Howard: Yes, I know that

Mark: You have more contacts with these other orgs to put toe in water
... OMG has put forward with a special interest group
... starting to bring awareness
... The case to developers stuck with UML

Karen: W3C is speaking with caBIG representatives
... have some IP questions

HOward: caBIG said they don't do triple stores
... in that area one of things that needs to happen is an active educational session
... run software to disprove myths

<nigam> Who are the caBIG representatives participating in the discussions?

Mark: What's fair to say...is that industry is looking at SemWeb

s/Howard/Harold

Harold: Stonebreaker

Q: was he going to column data bases

Masao: I am interested in this area
... What kind of engineer or researcher do you want to join from Asia
... drug person
... drug company person

<nigam> Q: Who are the caBIG representatives participating in the discussions?

Scott: Somebody who will bring a programmer with them
... so someone who can apply programming, since it's one of the bottlenecks
... I have been waiting for years for some things
... Need a clear idea of what they want to get; someone who has data

Eric: Need relevance

Scott: We can make extra teleconferences to try to work it out
... So RO1, in US
... from Mark

Mark: Europeans can apply, but harder
... Lots of ways people in this room and community could benefit from some of services we have
... raise the number of believers
... it's in everyone's interest
... NIH has not been favorable toward grants involving software engineering

<nigam> More at: http://www.bioontology.org/how-to-collaborate

Mark: One mechanism would be one of these collaborating projects

Kei: Ooes NIH fund community work?
... NSF does community based

Mark: It's unual
... are some special requests to create consortia, but those are rare

Scott: Along similar lines, EPI is interested in SemWeb approaches
... Eric and I talked to them about it
... about the gene atlas
... exposing as RDF

<matthias_samwald> s/EPI/EBI

Scott: This is a myth that people think going for SemWeb is a choice between RDF and Rel databases
... It's possible to do both an complement each other
... data warehousing and federation

Q: From a relational data base, you can flip back into an RDF representation on the fly

Scott: Yes, that's what Eric was there for
... he has written software to do that

Part of caBIG is caArray

scribe: I work with both groups
... Anatar
... From HTab
... Let's you produce your files
... caArray is interested in people accessing their cancer gene expression experiments

Scott: There were two people who attended caBIG in March

Harold: Gilberto Fragosio

Trish: Dave Howe

Scott: They were also there for shared names

Harold: Primary attraction for some
... original health care identifier morphed into HCLS

Josh: There is ongoing interest in caBIG community to apply SemWeb tech
... not make it an either or choice, but a complement

EricP: A certain amount of infrastructure you need to invest
... convince them that the "SemWeb hippies" will have some longevity

Jeff Abbott: I come from systems engineering background and human decision factors background

scribe: It's just hard to get people to change
... has to have a demo
... proof is in the pudding
... I would like to help with that
... In San Diego, they are fighting about whether to call themselves systems engineers or enterprise architects
... I would love to do a demo
... I am building an HSI template
... It's mandatory
... I'm building a template

Scott: SemWeb also has another myth
... that people think about having some giant cloud of ontology
... before you can do something
... Whereas a software engineering approach
... A guy from AstraZeneca, was talking about knowledge management
... he talked about policy level
... and how hard it was to share info inside the organization
... I asked if he would implement on a SemWeb basis
... and share outside
... He was for it
... A lot of people at that level don't think about that aspect
... SemWeb as an interoperability layer
... it's not obvious to everybody, esp. those thinking about policies

jeff: They may be thinking about other misgivings; you should ask them

Scott: In pharma, you need to have access control
... Eric has presented an approach
... control who can access what data
... by injecting user certificates into the query itself

Eric: Stick it next to part of query you want to restrict the access to

Scott: So if you could get hospital IT to use access control, they can control data and should be happy
... Still it's technically possible

Jeff: I did a project five years ago, people wanted to keep their data bases to themselves
... we had to do a lot of training and education
... It's a big obstacle
... Have to see what their concerns are

Scott: Another tact for engineers

Eric: Mechanical engineers like the explicit graph
... data model simplicity and publishing are clear

Scott: Any other ideas about resources
... I am strongly in favor of grants that would take members of this HCLS community along
... and give us more resources
... Some efforts to hire Eric that way, for example
... someone from W3C who can do more of that work
... beyond membership
... I hope there are other people who can come up with grant ideas and proposals to help move things along
... One evolved out of an ELS proposal

Kei: Yes, we need to pursue this
... It takes a long time
... nine months to a year
... before you get the funding

Harold: What about speakers from this group?

Scott: Susie, Eric and I do our best to spreak the work around
... We just visited EBI, Uniprot, BioIT
... Pharma IT, SemWeb PRISM Forum
... Everytime I talk about HCLS, there are people who are interested

Harold: So we can send invites
... good to know
... So let's take a break and be on schedule
... Yes, and I'll do my talk when we come back

<matthias_samwald> I need to leave the call now, bye everyone!

<ericP> thanks for coming matthias_samwald!

<Karen_> Scott: Ok, let's continue

<Karen_> Harold: So Semantic Wiki

<Karen_> ...much can be found on the ontolog site

<ericP> scribenic: Karen_

<Karen_> ..Focus on Semweb wiki

<Karen_> ...Here is BioMOBY on wikipedia

<Karen_> ...If it's red, it has not been created yet

<Karen_> ...So I can get invited in different ways

<Karen_> ...This is full of wonderful facts we would like to mine

<Karen_> ...but can only do it using some sort of MLP

<Karen_> ...Would be realy interesting to get that into triples

<Karen_> ...similarly

<Karen_> ...an entertaining problem

<Karen_> [looking for scroll bar, go to bottom]

<Karen_> ...Someone has assigned BioMOBY to a couple of categories

<Karen_> ...How this is done

<Karen_> ...Cut down HTML

<Karen_> ...link to something else inside

<Karen_> ...a registry of web services

<Karen_> ...which we link

<Karen_> ...someone put SemanticMoby in brackets

<Karen_> ...Go to bottom

<Karen_> ...and here are our the category links

<Karen_> ...The Semantic Wikipedia people

<Karen_> ...looked at this

<Karen_> ...and knows that BioMOBY belongs to category bioinformatics

<Karen_> ...They said BioMOBY is an instance

<Karen_> ...So we can glean the fact that Bioinformatics is a subclass of informatics

<Karen_> ...What's missing, to make it complete

<Karen_> ...Utilizes RDF

<Karen_> ...The verb is sitting out here

<Karen_> ...So we extend the syntax

<Karen_> ...and put your connector inside the brackets

<Karen_> ...now I have triple

<Karen_> ...BioMOBY...Utilizes...RDF

<Karen_> ...that is what they did

<Karen_> ...the addition of this to extend media wiki into semantic media wiki

<Karen_> ...So once you get semantic media wiki, I can edit pages

<Karen_> ...So I created this property

<Karen_> ...Scroll down to bottom, so when you hit a property, you can find relationships

<Karen_> ...And building something that looks like a triple store

<Karen_> ...What we are fighting to correct

<Karen_> ...It's so close to a triple store, it's frustrating that it's not [laughs]

<Karen_> ...The semantic media wiki people

<Karen_> ...believe that the text should be the source of the information

<Karen_> ...and the embedded info should be the objects

<Karen_> ...We are working with Ontoprise

<Karen_> ...We want a triple store behind this that goes both directions

<Karen_> ...So that's coming slowly

<Karen_> Eric: So I do a SPARQL update that creates a new protein

<allen> please provide "state of interest HCLS as a pdf file

<Karen_> ...and go to wiki page and it's sparsely populated

<Karen_> ...What I wrote into triple store

<Karen_> ...and read from wiki side, what do I see?

<Karen_> ...Looking for ideas

<Karen_> Harold: One thing you can do

<Karen_> ...is through a URL

<Karen_> ...but I'm doing it manually

<Karen_> ...I can pull what's recorded in wiki in RDF

<Karen_> ...You can pull stuff out in Jason

<Karen_> ...and use an Ajax server for other devices

<Karen_> ...By going to URLs

<Karen_> ...They have added

<Karen_> [Harold looking]

<Karen_> ...This should look familiar to people who know MIT Simile project

<Karen_> ...a neat way to do RDF data

<Karen_> ...You can select, slice and dice and organize as tables different view

<Karen_> ...of triple data

<Karen_> ...MIT did this

<Karen_> ...What's nice is this one has embedded a SPARQL test query

<Karen_> ...put in an ask and everything that has this relationship

<Karen_> ...with oncology lesion measurement

<Karen_> ...look at domain

<Karen_> ...look at formats

<Karen_> ...this is a SPARQL like query language

<Karen_> ...Allows you to construct pages and do fancy queries

<Karen_> ...Then become exportable in RDF and OWL

<Karen_> ...It's almost RDF

<Karen_> ..So almost an RDF end point

<Karen_> Scott: It's Exhibit

<Karen_> Harold: It builds itself almost like a mini triple store

<Karen_> ...So we have idea of creating, querying the triples

<Karen_> Nagim: This does not look like semantic media wiki

<Karen_> ...not embedded in some container

<Karen_> Harold: That's what's nice about media wiki

<Karen_> ...If there is something you want to do

<Karen_> ...somebody else has probably wanted to do it, too

<Karen_> ...So you select which solution fits your use case best

<Karen_> ...This is one of hundreds of little add-ons

<Karen_> ...Turned out to be useful

<Karen_> Nagim: But what I'm saying is that it's something else

<Karen_> Harold: no, it's semantic media wiki

<Karen_> ...This is one called category tree

<Karen_> ...showing that we loaded SNOMED CT

<Karen_> ...How we use it

<Karen_> ...the Ontolog people have created set of extensions called HALO

<Karen_> ...If I'm editing

<Karen_> ...mentioning properties

<Karen_> ...You can do an inline search

<Karen_> ...gene is a biomarker up

<Karen_> ...define properties

<Karen_> ...this one defined by NCI

<Karen_> ...This is another set of tools, the auto completion

<Karen_> Harold showing Naitonal Data Resource Catalog Prototype

<Karen_> ...We have created in such a form, we are creating 11179 records

<Karen_> ...loaded in as set of classes and relationship properties

<Karen_> Scott: If you can export RDF, how far is it away from exporting?

<Karen_> Harold: Would be good to convince semantic media wiki people

<Karen_> Scott: So this leads us to our next talk, SPARQL update

<Karen_> Harold: An earlier example, we used Prodigy to do a pull

<Karen_> ...and then fed back to wiki

<Karen_> Eric: I have lots of questions...for later

<Karen_> Nagim: So how well does this scale?

<Karen_> Harold: Very well

<Karen_> Nagim: A solution chokes out at about a hundred million

<Karen_> Harold: Media wiki has to built on relational

<Karen_> ...getting rest of community behind it

<Karen_> ...to start writing SPARQL queries on back end

<Karen_> ...If they are story queries in relational, just do the conversion

<Karen_> Harold: I think that's how Simile does it

<Karen_> Scott: Because that would scale

<Karen_> EricP: I would like to welcome the SPARQL Working Group

<Karen_> ...SPARQL is a spec that came out nearly two years ago

<Karen_> ...it is one of the primary RDF query languages now; it is a stable standard

<Karen_> ...First version didn't include some features

<Karen_> ...Some of those are coming around this time

<Karen_> ...What we have to do

<Karen_> ...Like "not exist"

<Karen_> [Eric stands in front of screen]

<Karen_> Eric explains chart of clinical trial patient in a study

<Karen_> ...write "optional, not found"

<Karen_> ...So if I want to find a patient and what type of medication he is taking

<Karen_> ...So like SQLs grouped by...

<Karen_> ...function that you can say, project out the some as some new variable total

<Karen_> Michael: If patient had two dates of birth?

<Karen_> ...SQL does not allow that

<Karen_> Lee: We discussed that you will not be allowed ot project out variables

<Karen_> Michael: DoB

<Karen_> Lee: Create in aggregate function

<Karen_> ...or if you wanted every DoB then include as part of group key

<Karen_> EricP: Sub queries

<Karen_> ...looking for some patient taking some substance administration

<Karen_> ...Some subselect where I'm finding their contra indications

<Karen_> ...that disqualify them

<Karen_> ...and I'm only taking one of them

<Karen_> Lee; Yes, exactly like sample

<Karen_> Eric: Concatonating

<Karen_> ...creating contraURL

<Karen_> ...Property paths

<Karen_> ...little star is complicated

<Karen_> [Karen invites another scribe, please]

<LenaDeus> ericP: [slide 9] using "Service" as a url to SPARQL

<LenaDeus> the exact mathematical operation is executed with data coming from the SPARQL services as the one used to answer the query being executed

<LenaDeus> Kei: how to issue a query to find a node with the most connectivity?

<LenaDeus> lee: group by protein and do the counting operation using aggregates

<LenaDeus> [slide 10]

<LenaDeus> graph management protocol

<LenaDeus> SPARQL update: just like a construct, but inserting into a graph

<LenaDeus> (same for delete)

<LenaDeus> service description: each service describes itself based on teh version of the sparql protocol

<LenaDeus> each service can also specify which functions are accepted (for example reasoning)

<LenaDeus> [slode 13]

<LenaDeus> D2R service description:

<LenaDeus> using PUT <request_uri> ... RDF payload ... would be equivalente to putting a graph in a certain URI

<LenaDeus> [slide 15] given 2 models (2 universes) where the inference leads to two different assertions

<LenaDeus> using blank nodes in the query, both viable models can be queried

<LenaDeus> (a query that operates over the logical intersection)

<LenaDeus> [slide 16] inference

<LenaDeus> inference should be used sparringly

<LenaDeus> [slide 17]

<LenaDeus> michael: XOR?

<LenaDeus> RIF: the idea was to cover a large number of rule languages such that the engine could be agnostic to the specifici rule language beign user

<LenaDeus> RIF was not desgined to be a universal rule language

<LenaDeus> RIF is n-ary

<LenaDeus> there is a syntax in RIF for triples

<LenaDeus> rif was designed to be extensible

<LenaDeus> RDFS allows type derivation; owl allows much more inference of types, including identity

<LenaDeus> in owl, there can be multiple represetnations of the same individual (even with different URI)

<LenaDeus> RDFS gives us "cheap" inference

<LenaDeus> these inferences do not validate

<LenaDeus> rdfs is a language that allows you to learn more things

<LenaDeus> however, it does not tell me that a mistake has been made

<LenaDeus> when owl is translated to rdf, it reasults in a long verbose syntax

<LenaDeus> [slide 27] there are therefore reasons why people will not want to query the RDF form of this information

<LenaDeus> owl 1.1 allows some syntactic sugar

<LenaDeus> owl 2 has lots of cardinality

<LenaDeus> annotation properties produce no inferences

<LenaDeus> punning: a property that belongs to a specific individual belonging to a class but does not belong to the class

<LenaDeus> POWDER: define a class of things

<Susie> let me know if you'd like me to set up livemeeting

<mscottm> yes, please!

<mscottm> We've got yesterday's code and it's not getting us in

<Susie> ok. it's set up.

<Susie> what code?

<ericP> just a sec

<ericP> trying to get zakim's attention

<mscottm> Susie - what is the code? Can you paste a link here?

<ericP> ARG!

<Susie> it's on the wiki

<mscottm> https://www.livemeeting.com/cc/jnjamericas/join?id=GH4DR6&role=attend&pw=jm-wR9S doesn't get us in

<Susie> https://www.livemeeting.com/cc/jnjamericas/join?id=GH4DR6&role=attend&pw=jm-wR9S

<mscottm> that looks like the same URL..

<ericP> Susie, 452571

<ericP> Susie, 42571

<ericP> topic HCLS F2F

HCLS F2F

<ericP> meeting: HCLS F2F

<Susie> try https://www.livemeeting.com/cc/jnjamericas/join?id=MPZK3R&role=attend&pw=r5*4%402P

<matthias_samwald> Zakim +danbri is matthias_samwald

<ericP> drop the '+'

<ericP> the leading '+' says the following identifier just joined

<ericP> and plus your point

<ericP> matthias_samwald, i had to mute you

<matthias_samwald> fixed it..

<matthias_samwald> noise gone?

<mscottm> yes

<matthias_samwald> thanks

<mscottm> https://www.livemeeting.com/cc/jnjamericas/join?id=MPZK3R&role=attend&pw=r5*4%402P

<ericP> ahh, you already had

<ericP> sorry

<mscottm> Susie - Allen_Hobbs has Peter's presentation for LiveMeeting

<mscottm> Is Allen_Hobbs the one 'in control' of the screen in LiveMeeting now?

<Susie> I've made allen presenter

<Susie> he needs to share his slides

<ericP> Susie, conf code is 42571

<Susie> stuck on another line...

<ericP> okidoke

<Susie> i can see allen's desktop

<ericP> sweet!

Applications of SNOMED and DL in Kaiser's EHR [Peter Hendler, Kaiser Permanente ]

<ericP> [slide Three Layered Terminology]

<ericP> Peter: map all our terms to SNOMED

<ericP> ... also to ICD and CPT when needed for billing

<ericP> ... CPT only talks about how much work something is

<ericP> ... e.g. "repair of wound 1-cm"

<ericP> ... not medically relevent

<ericP> [slide Example Interface]

<ericP> [slide SNOMED is EL+]

<ericP> [slide Pulminary tuberculosis]

<ericP> scribenic: ericP

<ericP> [slide Pulminary tuberculosis SNOMED Multiple Hierarchies]

<LenaDeus> http://www.cliniclue.com/

<ericP> [slide Pulminary TB]

<ericP> [slide Query Model]

<ericP> [slide Structural Subsumption]

<ericP> scribenic: mscottm

<mscottm> Hi Elgar - we are live on LiveMeeting (link at bottom of F2F page)

<mscottm> Comparison is performed after 'normalization'. You can see if a term is subsumed by another if they are normalized.

<mscottm> Peter: But Kaiser needs negation and disjunction

<mscottm> .. once you add negation and/or discuntion to an ontology, you can't perform structural subsumption anymore. You need tableau reasoning

<ericP> [slide Applications of SNOMED and DL in Kaiser's EHR]

<ericP> oops

<ericP> [slide A 64 bit machine ...]

<ericP> Peter Hendler's SNOMED computer

<mscottm> ..created an Amazon EC2 image that could hold SNOMED (64bit, 12G RAM, etc.)

<mscottm> ..tested with some classifications and it seemed to provide reasonable results

<mscottm> ..tested classification with negation (example: non-infectious pneumonia)

<mscottm> ..used cliniclue to see what the codes in Protege meant

<mscottm> ..What happens if you put in lots of negation and disjunction? Massive ammounts - 25000 anatomical terms made into negated (ala non-infectious) terms

<mscottm> ..made 25000 disjunctive terms from "malignant or infectious disorder of"

<mscottm> ..Protege with Fact reasoner took about 37 minutes without neg/disj, 1 hour and X min *with*

<mscottm> ..tried with OWL API and it took 35 min

<mscottm> ..for some reason Protege with Pellet didn't seem to work (fast enough)

<nigam> nope .. don't have car

<nigam> btw, how do you send the kind of message you just did (i.e. direct to me)

<nigam> Webex for BioRDF breakout is started

<colin> is there anybody there?

<matthias_samwald> yes?

<matthias_samwald> they are still in the first session

<colin> mscottm, nobody there, other than me and EricP

<nigam> "/me" mcottm, I just joined there as well

<nigam> "/nigam"

<mscottm> Hi All, we should join HCLS1 for BioRDF breakout and HCLS2 for LODD/TMO breakout

<nigam> Kei you now have control of the webex

<matthias_samwald> still starting up webex.

<matthias_samwald> but feel free to start

<ericP> jun, ears on?

<nigam> please use HCLS1 line for audio .. do not use webex

<nigam> "/ericP" no, only jun seems to be on the webex audio ..

<matthias_samwald> i am on the wrong web ex. at the moment, i see allen hobbs checking his e-mails.

<matthias_samwald> !!!

<matthias_samwald> could someone please give me the correct webex address AND tell the other person that he is transmitting his desktop via webex?

<matthias_samwald> (i didn't look closely! ;) )

<nigam> Correct webex = https://stanford.webex.com/stanford/j.php?ED=123758442&UID=1139021767&PW=NMDAwNDQyZTVl&RT=MiM0

<mscottm> Matthias - you're looking at the LiveMeeting link for Peter Hendler's presentation

<mscottm> scribenick: mscottm

Michael: Just for the record, MIAME isn't a format but description of the information to be delivered (in a format)
... MAGE-TAB specifies a set of files associated with the experiment
... Investigation Description Format (IDF) - a sort of summary of investigation info
... Array Design Format (ADF) - tab-delimited file defining each array type used, design of the array, probe sequences.
... Sample and Data Relationship Format (SDRF) - tab-delim file describing relationships betw samples, arrays, data and other objects used in the investigation, and misc MIAME info catchall.
... Raw and processed data files - ASCII or binary files, typically in native format. Alternatively, can be provided in tab-delim.
... [going through IDF example]
... [discussing contents of MAGE-TABv1.0.pdf]
... [Investigation design graph diagram on page 5]
... [how different types of experiments (DAG) are represented: dual channel, single channel, biological replicates, time series, ..]

Scott: I recently became aware of ISATAB format as a superset of MAGE-TAB

Michael: Yes, ISATAB is also a contender.

<matthias_samwald> http://hcls.deri.org/atag/

<ericP> 7tm query over aTags

<nigam> How hard would it be to suggest aTags using ontologies in NCBO

<matthias_samwald> redialing...

<nigam> Can I close the Webex?

<matthias_samwald> hi everyone, i need to leave now. have a good time for the rest of the meeting!

<matthias_samwald> (i will still watch irc now and then)

<nigam> I am leaving as well! .. IRC is logged in and will check in occasionally. Have a good meeting!

http://esw.w3.org/topic/HCLSIG/Meetings/2009-11-02_F2F

WebEx (NCBO hosted) URL: https://stanford.webex.com/stanford/j.php?ED=123758442&UID=1139021767&PW=NMDAwNDQyZTVl&RT=MiM0 , Enter your name and email address., Enter the meeting password: ncbomeeting, Click "Join Now".

Elgar is from AstraZeneca (not Pfizer).

<Roger> Are these presentations available somewhere?

<ericP> F2F agenda (with presentation links)

<Roger> Thanks.

<Roger> This sucker is 8 MB!

<ericP> Superman problem