Multi-stakeholder Perspectives on Translational Medicine

Resources: <SWObjects sparql binary> and <sample queries>
(contains <goProt.map> <goProt.rq>. <goProt2.map> <goProt2.rq> <goProt2-bug.rq>.)

Eric Prud'hommeaux, Sanitation Engineer. Helena Deus, Life Sciences Geek.
Last modified: $Date: 2011/02/23 16:18:33 $

This work is licensed under a Creative Commons Attribution 3.0 License, with attribution to W3C.

Agenda:

Intro to W3C,HCLS.
Health care, life science landscape.
Applicability of RDF.
Translational Medicine Ontology.
Linked Open Drug Data
RDF technology survey.
SWObjects Tutorial?

who is w3?

~420 member companies, universities, govornment agencies
started as "HTML consortium"
focus on architectural issues:
- protocols
- information representation
- accessibility
- internationalization
- device independence
good success rate
google says we're relevent

who is HCLS?

six task forces:

Terminology – SemWeb representation of existing resources
Scientific Discourse – building communities through networking
BioRDF – integrated neuroscience knowledge base
Clinical Observations Interoperability – patient recruitment
Linking Open Drug Data – aggregation of Web-based drug data
Translational Medicine Ontology – patient-centric upper level

Other Projects: Clinical Decision Support...

HCLS Mission

Life science complexity grows exponentially.
Analysis requires increasing domain interaction.

We develop RDF tools, data and practices to help academia/industry use RDF.

Health Care Landscape:

Hospital databases.
Clinical trials results.
Personally-controled records.

Hospital data:

Hand generated.
De-identified database.
De-identified RDF dumps.

Indivo

PCHR system.
XML-serialization.
Open source.

Easy for us to extend schemas and feed back to project.

Easy for TMO's medical vocabs to become normative.

Nascent convergence of needs:

Clincal data custodians.
Translational researchers.
Patients.
Public health workers.
...

RDF's Fortés

Common data model.
Web scale resource distribution.
Simple interpretations of "tag soup".
Decentralized extensibility.

Effects:

Latent integration
Lightweight "standards".

Decentralized Extensibility

Unambiguous identifiers for things, types and properties.
Culture of re-use.
Decentralized invention.
Principled Architecture

Decentralized Extensibility

Unambiguous identifiers for things, types and properties.
Culture of re-use.
Decentralized invention.
Principled Architecture

Extensibility Example: Blood Pressure

@prefix edns: <http://www.loa-cnr.it/ontologies/ExtendedDnS.owl#> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix galen: <http://www.co-ode.org/ontologies/galen#> .
@prefix snomed: <http://termhost.example/SNOMED/> .

<encounter7> edns:patient <patient3> ;
             edns:screeningBP <s20090714c> .
<patient3>   r:type galen:Patient ;
             foaf:family_name "Levin" ;
             foaf:firstName "Henry" .
<s20090714c> dc:date "2009-07-14T18:23"^^xsd:dateTime ;
             edns:systolic "132"^^edns:mmHg ;
             edns:diastolic "86"^^edns:mmHg .

Extensibility Example: Blood Pressure

@prefix edns: <http://www.loa-cnr.it/ontologies/ExtendedDnS.owl#> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix galen: <http://www.co-ode.org/ontologies/galen#> .
@prefix snomed: <http://termhost.example/SNOMED/> .

<encounter7> edns:patient <patient3> ;
             edns:screeningBP <s20090714c> .
<patient3>   r:type galen:Patient ;
             foaf:family_name "Levin" ;
             foaf:firstName "Henry" .
<s20090714c> dc:date "2009-07-14T18:23"^^xsd:dateTime ;
             edns:systolic "132"^^edns:mmHg ;
             edns:diastolic "86"^^edns:mmHg ;
             edns:posture snomed:_163035008 . # SNOMED:sitting

Emergency Room Blood Pressure Query

PREFIX edns: <http://www.loa-cnr.it/ontologies/ExtendedDnS.owl#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX galen: <http://www.co-ode.org/ontologies/galen#>
PREFIX snomed: <http://termhost.example/SNOMED/>

SELECT ?date ?systolic ?diastolic WHERE {

?encounter edns:patient     ?patient ;
           edns:screeningBP ?screenBP .
?patient   foaf:family_name "Levin" ;
           foaf:firstName   "Henry" .
?screenBP  dc:date          ?date ;
           edns:systolic    ?systolic ;
           edns:diastolic   ?diastolic ;
} ORDER DESC BY ?date

Clinical Study Blood Pressure Query

PREFIX edns: <http://www.loa-cnr.it/ontologies/ExtendedDnS.owl#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX galen: <http://www.co-ode.org/ontologies/galen#>
PREFIX snomed: <http://termhost.example/SNOMED/>

SELECT ?date ?systolic ?diastolic WHERE {

?encounter edns:patient     ?patient ;
           edns:screeningBP ?screenBP .
?patient   foaf:family_name "Levin" ;
           foaf:firstName   "Henry" .
?screenBP  dc:date          ?date ;
           edns:systolic    ?systolic ;
           edns:diastolic   ?diastolic ;
           edns:posture ?posture . 
FILTER (?posture = snomed:_163035008 || ?posture = snomed:_163035009)
} ORDER DESC BY ?date

Clinical Study Blood Pressure Query

PREFIX edns: <http://www.loa-cnr.it/ontologies/ExtendedDnS.owl#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX galen: <http://www.co-ode.org/ontologies/galen#>
PREFIX snomed: <http://termhost.example/SNOMED/>

SELECT ?date ?systolic ?diastolic ?posture WHERE {

?encounter edns:patient     ?patient ;
           edns:screeningBP ?screenBP .
?patient   foaf:family_name "Levin" ;
           foaf:firstName   "Henry" .
?screenBP  dc:date          ?date ;
           edns:systolic    ?systolic ;
           edns:diastolic   ?diastolic .
  OPTIONAL {
    ?screenBP  edns:posture ?posture .
  }
} ORDER DESC BY ?date

Many simple domains ...

... in one universe.

Global continuity

Connect my data to other known data.
Collaborative workspace.

Collaborative workspace

Reflects our desire to be lazy/effective/efficient.
Reflects the reality of workflow complexities .

Heterogeneous access to

Public databases.
Private/institutional data.
Data from limited partnerships.

Continuity of data ...

... enables mashups to become queries.

SERVICE <http://uu3.org:8888/7tm_receptors>
  {
    ?iuphar  iface:family      ?family .
    ?iuphar  iface:code	       ?code .
    ?iuphar  iface:iupharName  ?iupharNm .
    ?human   iface:iuphar      ?iuphar .
    ?human   iface:geneName    "GABBR1" .
    ?human   iface:entrezGene  ?humanEntrez .
  }
SERVICE <http://dbpedia.org/sparql?default-graph-uri=http%3A%2F%2Fdbpedia.org&should-sponge=&>

  {
     _:gene  dbp:entrezgene    ?humanEntrez ;
             rdfs:label        ?label ;
    FILTER (lang(?label) = "en")
  }
GRAPH <http://hcls.deri.org/atag/data/gabab_example.html>
  {
     ?topic  rdfs:label        ?label .
     ?post   sioc:topic        ?topic
  }

Translational Medicine Ontology

Many ontologies
(e.g. patients, biomarkers).
Managed by local concensus.
Provides connective tissue.
Development methodology.

HCLS TMO, who needs what?

Category	User / Role	Interest
Research	Biologist (in vivo, in vitro, cellular & molecular)	Target identification, assay development, target validation
	Bioinformatician	Biological knowledge management, cellular modeling
	Immunologist	Natural defense mechanisms
	Cheminformatician	Predictive chemistry
	Medicinal chemist	Drug efficacy
	Systems physiologist	Tolerance, adverse events
Clinic	Clinical trial specialist	Trial formulation, recruitment
	Clinical decision support	Data Analysis, trend finding
	Primary care physician	General, conventional care
	Specialty medical provider	Specialized treatments
Business	Sales & marketing	Revenue generation
	Strategic/portfolio manager	Assessing market opportunities
	Project manager	Prioritizing resources & activities
	Health plan provider	Insurance coverage

AD patients without the APOE4 allele?

SELECT DISTINCT ?name ?patient
WHERE {
 graph <pchr_new> {
  ?patient 
   foaf:name ?name;
   trans:hasCondition [ trans:diagnosedWith trans:alzheimers_disease ] .
   FILTER ( 
    !bif:exists
     (( 
       SELECT ?result
       WHERE {
        ?encounter 
        trans:patient ?patient ;
        trans:test [ a ?testname ;  trans:result ?result ] .
        ?result
         trans:feature trans:variant_APOE4 ;
         trans:present true . 
       }      ))     )  } }

results

Linked Open Drug Data

8 million triples
370,000 links
10 datasets
DrugBank LinkedCT DailyMed DBpedia Diseasome RDF-TCM RxNorm SIDER STITCH Medicare ChEMBL
that we know of

drug knowledge

  @prefix : <http://rdf.farmbio.uu.se/chembl/onto/#> .
  act:a108377 :onAssay ass:a50464 . ass:a50464 :hasTarget p:t53 . p:t53 :hasKeyword "Isomerase" .
  act:a108377 :extractedFrom rsc:r6368 . rsc:r6368 bibo:pmid	pubmed:15125947 .
  act:a108377 :forMolecule mol:m258506 . mol:m258506 bo:smiles "COC(=O)[C@@H]1C[C@@H](CN1)NCc2n[nH]c3cccc(OCc4ccc(cc4)C(C)(C)C)c23" .
  act:a108377 :standardUnits "ug mL-1" .

Exploits LOD:

Schema discovery.
Scales like the Web.
Familiar maintenance.
Confidence of interpretation.

LODD Resources

RDF Channeling Existing Data

XML
Relational Databases
Spreadsheets
NLP

XML data

<?xml version="1.0"?>
<ClinicalDocument transformation="hl7-rim-to-pomr.xslt">
  <recordTarget>
    <patientRole>
      <patientPatient>
	<name>
	  <given>Henry</given>
	  <family>Levin</family>
	</name>
	<administrativeGenderCode code="M"/>
	<birthTime value="19320924"/>
      </patientPatient>
    </patientRole>
  </recordTarget>
  <component>
    <StructuredBody>
      <Observation>
	<code displayName="Cuff blood pressure"/>
	<effectiveTime value="200004071430"/>
	<targetSiteCode displayName="Left arm"/>
	<entryRelationship typeCode="COMP">
	  <Observation>
	    <effectiveTime value="200004071530"/>
	    <value value="132" unit="mm[Hg]"/>
	  </Observation>
	</entryRelationship>
      </Observation>
      <Observation>
	<code displayName="Cuff blood pressure"/>
	<effectiveTime value="200004071530"/>
	<targetSiteCode displayName="Left arm"/>
	<entryRelationship typeCode="COMP">
	  <Observation>
	    <code displayName="Systolic BP"/>
	    <effectiveTime value="200004071530"/>
	    <value value="135" unit="mm[Hg]"/>
	  </Observation>
	</entryRelationship>
	<entryRelationship typeCode="COMP">
	  <Observation>
	    <code displayName="Diastolic BP"/>
	    <effectiveTime value="200004071530"/>
	    <value value="88" unit="mm[Hg]"/>
	  </Observation>
	</entryRelationship>
      </Observation>
    </StructuredBody>
  </component>
</ClinicalDocument>

Patient history
- Name
- Age
- Illnesses
Encounter details
- Reason for visit
- Vitals (e.g. BP)
- Tests
- Prescriptions
Facility details…

Mapped to RDF

RDF representation of XML

Indivo developers see forms.
Indivo developers see XML.
We see RDF (via GRDDL).

Relational data (e.g. EPIC, I2B2) …

… as a graph.

Using Existing Data(bases)

Global names.
http://someheathprovider.example/patientVisits/C1#info
Simple, declarative sentences.
Global links.
Typed values.

Using Existing Data(bases)

Global names.
http://someheathprovider.example/patientVisits/C1#info
Simple, declarative sentences.
Global links.
Typed values.

Using Existing Data(bases)

Global names.
Simple, declarative sentences.
provider:C1 role:patient provider:P1 .
Global links.
Typed values.

Using Existing Data(bases)

Global names.
Simple, declarative sentences.
provider:C1 role:patient provider:P1 .
Global links.
Typed values.

Using Existing Data(bases)

Global names.
Simple, declarative sentences.
Global links.
http://someheathprovider.example/roles#patient
Typed values.

Using Existing Data(bases)

Global names.
Simple, declarative sentences.
Global links.
Typed values.
e.g. ISO 8601 datetimes
provider:P1 role:birthdate "1946-11-03"^^xsd:date .

Relational Mappings

D2R
Relational Views
RDF Views

D2R

<d2r:ClassMap
   d2r:type="eb:Customer"
   d2r:sql="SELECT Tab_K_Kunde.KundenNr, Tab_K_Kunde.FullName,
        Tab_K_Kunde.Anrede, Tab_K_Kunde_Kreditkarte.CCNo
        FROM Tab_K_Kunde_Kreditkarte INNER JOIN Tab_K_Kunde ON
        Tab_K_Kunde_Kreditkarte.KundenNr = Tab_K_Kunde.KundenNr;"
   d2r:groupBy="Tab_K_Kunde.KundenNr"
   d2r:uriPattern=
       “http://example.org/customers#@@Tab_K_Kunde.KundenNr@@“>

Re-uses OWL constructs:

<d2r:ObjectPropertyBridge d2r:property="rdf:type"
   d2r:pattern="iswc:@@Persons.Position@@"/>

Configuration consumable as RDF (SPARQL-able).

Relational Views (DDL)

create iri class prd:product_iri using
  function OPLWEB.DBA.PRODUCT_URI (in id varchar)
    returns varchar,
  function OPLWEB.DBA.PRODUCT_URI_INVERSE (in id_iri varchar)
    returns varchar .

Uses hardened algorithms from SQL land.

RDF Views

   PREFIX :mydb <http://cityhospital.example/dbs>
CONSTRUCT { ?o a               study:SubjectObservation .
            ?o study:subject   ?p .
            ?o study:clinician ?d .
            ?d :foaf:name ?dName }

    WHERE { ?o mydb:patient ?p .
            ?o mydb:doctor  ?d .
            ?d mydb:name    ?dName }

Intuitive (testable) semantics.

RDB2RDF Status

Inference/"Validation"

(a,b,c) set enumeration

union

disjunction

algebraics

intersection

complement

restriction

cardinality

equivalence

courtesy: Fabien Gandon

Summary

Health care/life sciences a proving ground for the Semantic Web.
RDF tackles a lot of the requirements.
TMO a useful connective tissue.
Which RDF tools do what.

SWObjects Tutorial?

Acknowledgements

Nigam Shah
M. Scott Marshall
Helena Deus
HCLS IG gang
RDB2RDF WG gang
Sugar Hill gang

Questions?

Surely there are none.