1. The SOFG Anatomy Entry List - SAEL
Helen Parkinson, EBI
On behalf of the Standards and Ontologies for Functional Genomics SAEL Working Group
EBI, centre for bioinformatics, European and worldwide links for microarray data

2. Some History
SOFG 1 Hinxton 2002, Anatomists had a breakout group to discuss integration of Anatomy ontologies.
Outcome website set up listing known anatomy resources and ‘view’, and ‘intent to integrate expressed’

3. Gratuitous Advertising – SOFG2

4. SOFG/SAEL Workshop
David Shotton – “what if the ontologies are not orthogonal?”
In April 2004 an international workshop was held in Edinburgh to consider issues raised by the SOFG discussion
Participants representing users of Anatomy Ontologies
ArrayExpress, RAD, GXD, EMAGE
Participants representing Anatomy Ontologies:
Foundational Model of Anatomy (FMA)
GALEN, Mouse Adult Anatomy
Mouse Developmental Anatomy (EMAP)
Edinburgh Human Developmental Anatomy
CBIL Controlled Vocabulary for Anatomy

5. Integration issues Tissues Things made of the tissues
Cells making up the tissues (scale)
Correspondences Homologies/Orthologies – mouse tail/C.elegans tail
Stages
Developmental derivation
Relationship types, part-of, is-a, etc and how these are used differently
Considered a specific use case:
The use of anatomy ontologies in the functional genomics domain
Slide: Alan Rector, Jeremy Rogers

6. Functional Genomics Experiments
Sample-based functional genomics experiments are usually limited to what is obtainable by conventional dissection
High level terms are useful
Detailed ontologies are also useful where experimenters need these, for example when laser capturing samples
There are excellent resources already
Truth: Biologists are resistant to data sharing, annotation, standardisation, ….

7. Standard terms are needed for querying

8. MGED Ontology
Supports MIAME, provides terms for annotation of experiments (where they do not exist externally)
Creates a framework to reference external ontologies – therefore we need external resources
Requires that external terms be identifiable
Is implemented in data capture applications
An anatomy list for this domain needs to be simple and be flexible
Annotation needs are diverse
Multiple resources can be confusing

9. A multiplicity of resources
Many: resources, formats, philosophies, purposes, variable content,
Ontology
Accessibility
Content
CBIL Controlled Vocab. For Anatomy
CBIL browser
~600
Adult Mouse Anatomy
DAG-Edit/Jax Viewer
~2400
EMAP- Mouse development
DAG-Edit, Anatomy Browser
~8000
GALEN (Anatomy Only)
Protégé, Demo-GCE or OpenKnoME
~10,000
FMA
FM explorer
70,000 concepts >110,000 terms
Compare the FMA vs. the adult mouse

10. Anatomy Terminologies and Ontologies
NCI
MeSH
GALEN
SNOMED
Jax
is-a
Hollow viscus (body structure
Heart structure (body structure)
Entire thoracic viscus (body structure)
TubularCardiovascularComponent
IntrathoracicCardiovascularComponent
IntraMediastinalStructure
Thoracic cavity structure
FMA
Organ with cavitated organ parts
Body Organ
Cardiac Structure
Cardiopulmonary System
Cardiovascular System
Cardiovascular
System
-part-
Slide: Cornelius Rosse

11. Foundational Model of Anatomy (FMA)
FMA uses a frame-based formalism
concerned with the representation of concepts and relationships in a form that is understandable to humans and machine readable
Human/vertebrate
Definition: structural attributes
Content: organism to biological macromolecule
Serves as a reference ontology
Slide: Cornelius Rosse

12. FM Explorer

13. Adult Mouse @ Jax Vocab browser
Anatomical structures are organized spatially and functionally, using 'is a' and 'part of' relationships
For TS28
Purpose, encoding and integration of mouse gene expression data

14. SAEL .. is a simple list of ~120 terms
is for low-resolution descriptions of sample origin
terms have ids: SAEL:1
contains vertebrate terms at present
is NOT a new anatomy ontology
does NOT have defined relationship types
terms do not have definitions
is a first step to considering the relationships between the SAEL source ontologies
is NOT intended to replace deeper integration efforts

15. The SAEL current version 1.0
Download from
In plain text/OBO format
Will be maintained by MGED Ontology working gp + Terry Hayamizu (Jax)
Suggestions through MGED Ontology sourceforge tracker
Report on the workshop is available
Review publication from this workshop in CFG

16. Testing the content SAEL maps to ~ 80% of current terms tested
So far been tested vs. current annotation in
ArrayExpress/MIAMExpress – OrganismPart –82 terms
HGMP – microarray mouse and human only –97 terms
SMD - free text microarray sample annotations –22 terms
GXD - * 80% for blot and cDNA data only
RAD – microarray, uses CBIL

17. Implementation MIAMExpress – ArrayExpress data capture tool uses SAEL
Data in ArrayExpress will be mapped to SAEL
Future submissions will use SAEL
We will encourage users of the MGED ontology to use SAEL where appropriate

18. COBrA
XSPAN demo takes place on Wednesday, August 4th at 11.30am in room Alsh 2 – Albert Burger

19. Mapping source ontologies to SAEL
Adult mouse anatomy, FMA mapped to date
Mouse Developmental, GALEN, CBIL + others to do
Using COBrA from the XSPAN project -
Allows manual mapping between ontologies
Creates an OWL format mapping file
Reads: DAGEdit flat file format, GO XML/RDF, GO RDFS and OWL
Mappings available from

20. Web services A WSDL has been defined for SAEL:
dev_stage, is_tissue, is_cell_type, is_organ, is_system, superclass, subclass, part, part_of, uri, definition, authority, history, name, synonym
Source ontologies will provide a web service supporting queries and returning the attribute list
WSDL has been tested vs. Adult mouse anatomy, FMA and developmental mouse anatomy, will be tested further

21. Proposed web services architecture
SAEL and will be made available via user interface and programmatically
Querying multiple ontologies will be supported by the central SAEL web service
The SAEL Portal provides a graphical user interface for researchers to look up the mappings between the SAEL list of anatomical entities and the target ontologies.
Will be implemented in 2 phases, SAEL portal first, then local ws

22. Future
We welcome collaboration and mapping of other relevant ontologies proceeds e.g. EVOC
Building web services architecture, XSPAN
Refining SAEL, completing mapping
Inclusion in MGED Extended ontology (v1.2)
Deeper integration of anatomy ontologies
Decisions on handling of null mappings
Modification of relationship types in current version
Protégé version from Alan Rector

23. Acknowledgements
SAEL workshop participants: Stuart Aitken, Albert Burger, Richard Baldock, Jonathan Bard, Duncan Davidson, Terry Hayamizu, Helen Parkinson, Alan Rector, Jeremy Rogers, Martin Ringwald, Cornelius Rosse, Chris Stoeckert
John Gennari
Niran Abeygunawardena, EBI Website, MIAMExpress implementation
Funders: MRC-HGU, MGED, EU – TEMBLOR
Jeremy Gollub -SMD, Naran Hirani/Tom Freeman – HGMP

24. Gratuitous Advertising – SOFG2

25. Bio-Ontologies Panel Discussion
Michael Ashburner, Dept of Genetics Univeristy of Cambridge
Crispin Miller, Bioinformatics and Onco-informatics Group
Jeremy Rogers, Medical Informatics Group, University of Manchester
Barry Smith, Institute for Formal Ontology and Medical Information Science, University of Leipzig and Buffalo, State University of NY

26. SWOT analysis
131,000 hits on Google

27. SWOT 2 Strengths Weaknesses Opportunities Threats
Orientated towards the internal aspects of bio-ontologies or an individual ontology
Opportunities
Threats
Those factors external to bio-ontologies or an individual ontology

28. Panel Perspective Michael Ashburner Jeremy Rogers Crispin Miller
“Pragmatic ontologies for real utility in biology ”
Jeremy Rogers
“Pragmatic economic or user-led development risks the lowest common denominator or the mediocre: the Trabant or the Ford Mondeo. Theory-led development may ignore practicality and expense: the Formula One race car. How can such extremes be avoided in ontology engineering ? ”
Crispin Miller
“There is generally a difference between what we would like to say, and what our computers are capable of interpreting. How do we build ontology-based systems that can successfully resolve the tensions arising from this ”
Barry Smith
“Physics has pure mathematics as its formal backbone. What is the counterpart of pure mathematics in biology? Answer: formal ontology. ”

29. Criteria for future chairs
“combination of School Mistress sternness, eloquence, and opinionation” Phil Lord
“ability to pick people out of the audience” Robert Stevens

30. Michael GO
Strengths:wide uptake, community project, designed for a single problem – gene product attributes, dev of ontol pragmatic (weakness?), open, to sw
Weakness – Pragmatic design and build, qc mechanism, and implementation issues in DB, lack of formalism, no idea would be universal
Opp. - world domination, achieving greater integration across species
Threats – long term stability – academia, funding models, diversion into philoso

31. Crispin S - structure info W - O – abstraction level problematic
T - Knowing where to stop, ontology is linked to tools, cultural issues,

32. Jeremy Strengths: community of eager users – scope, where to start
Open source
T – semantic web hype, succession of curators

33. Helen S – collaboration
W – legacy data management, costs, maintenance, user uptake
O – improved data retrieval, query, formalisation,
T – reinvention of the wheel

34. Barry Smith MA – “completion/failure” S – exists
O – data, make it interoperable
W – pragmatic decisions become entrenched,
T – fools paradise – OWL is not expressive enough
expressive power – ‘cheating’ is-a overloading, need a top level of ontology – philosophical qu. way that instance is used
Ontology – reality – knowledge – describing knowledge, - assay ontology ? Ontology of scientific expts
Perfection is the enemy of the good !
Solution: ‘more precison’