Slide 1

Download Report

Transcript Slide 1 

the cancergrid metadata registry and toolset
University of Cambridge: Irene Papatheodorou, Lorna Morris; James Brenton.
University of Oxford: Charles Crichton; Andrew Tsui; Steve Harris; Jim Davies; Jeremy Gibbons.
reasoning
the software
• workgroup-scale ISO11179 metadata registry
• free BSD license
• open standards implementation: eXist; XSLT; XQuery; SKOS; RDF;
XForms; XML Schema; WSDL
• platform independent: five minute install
• transforms caDSR shopping cart content
• simple data element registration
• user classification through SKOS/Protégé
smart documents
We’ve connected our metadata registry,
our lexBig servers, the EVS and caDSR to
a query service manager giving a
uniform query interface to these
knowledge resources that may be
accessed from both Java and .NET. You
can use this service to:
• compose new UML models out of CDEs
and register new CDEs in Enterprise
Architect without leaving the class
diagram
• ‘type a column range’ in MS Excel
according to a CDE, naming the column,
controlling cell display and creating a
new metadata sheet with CDE
definitions persisted alongside your data
• use CDE and concept references to
design case report forms in MS InfoPath
• allow the user to access terminology
resources when completing InfoPath
based CRFs (with the VA)
cgMDR
lexBIG
caDSR
EVS
SAWSDL annotations of schemas and WSDL
documents support reasoning about content
• SQIV/XQuery uplifts the schema data model
into appropriate RDF
• assertions from SKOS classification schemes
or OWL ontologies classify elements in the
data model according to the type of data and
the type of user
• SPARQL/DL Queries can answer simple
questions reliably and scalably:
• does the data contain personal
information?
• which elements contain personal data?
• is the data suitable for the user
requesting the data?
• XML elements/RDF triples can be
replaced/blurred according to the results of
the queries
query service manager
(web service)
transformation
with the METABRIC project
•pathology data collected from UK and Canada
• metadata defined for each data type
• metadata defined for the transformation
target with common semantics
• SPARQL rules defined for the transformation
• combined dataset used for tissue analysis
• similar transformation work done in XSLT
• working through issues of registration of
transformation functions
CDE: histological type
of breast tumour Guy’s
Enumerated Value Domain:
1 In-situ ductal only
10 Tubular/Cribiform
11 Ductal Grade Unknown
12 Mixed…
CDE: histological type
of breast tumour
Manitoba
Enumerated Value Domain:
1 Ductal NOS
10 Ductal and Spec
11 Invasive tumour
12 DCIS…
CDE: histological type
of breast tumour
Nottingham
Enumerated Value Domain:
1 Inv. ductal/no spec type
10 Tubular
11 Mucinous
12 Invasive cribriform…
CDE: histological type
of breast tumour
Addenbrookes
Enumerated Value Domain:
A None
D DCIS
ID Invasive ductal
ID-IL …
CDE: histological type
of breast tumour
Vancouver
Enumerated Value Domain:
1 Carcinoma, NOS
10 Tubular adenocarcinoma
11 Merkel Cell Carcinoma
12 Papillary adenocarci…
object class: breast
tumour sample
metadata mining
property: assessment
with the VA
Xforms Post generation customisation
Generated view element
‘mining’ metadata out of CRFs from a SharePoint CTMS.
• custom program annotates schema elements with form text
• XML Schema uplifted into MDR, creating names and identifiers
• metadata elements
curated/rationalised
• ‘standard’ data
elements re-create
schemas with SAWSDL
annotations
• annotated schemas
create registered
forms
<xforms:input
ref="//cgMDR:change_description">
<xforms:label>Change Description</xforms:label>
</xforms:input>
Transformation
<xsl:template match=
"xforms:input[ends-with(./@ref,'cgMDR:change_description')]">
<xforms:textarea
ref="//cgMDR:change_description">
<xforms:label>Change Description </xforms:label>
</xforms:textarea>
</xsl:template>
Customised view element
<xforms:textarea
ref="//cgMDR:change_description">
<xforms:label>Change Description</xforms:label>
</xforms:textarea>
data element concept:
histological type of
breast tumour
Jena rules for transformation
CDE: histological type of
breast tumour METABRIC metaanalysis
Enumerated Value Domain:
common semantics
conceptual domain:
breast tumour histology