Hinxton meeting March 2008
Attendees
- Michael Ashburner, Cambridge University
- Johnathan Bard, Edinburgh University
- Bernard de Bono, EBI
- Thorsten Henrich, EMBL
- Suzi Lewis, LBL
- Jane Lomax, EBI
- James Malone, EBI
- David Osumi-Sutherland, Cambridge University
- Helen Parkinson, EBI
- Monte Westerfield, Zfin
Agenda
- Johnathan Bard - Minimal Anatomy Ontology
- Thorsten Henrich - The Bilateria Ontology
- David Osumi-Sutherland
http://www.bioontology.org/wiki/index.php/Image:DOS_caro-MRCA_080321.ppt
- Helen Parkinson
http://www.bioontology.org/wiki/index.php/Image:Caro_hinxton_2008.ppt#filelinks
- Bernard de Bono
Discussion
The presentations focussed on:
- the need and use cases for a cross species anatomy ontology for annotation and analysis purposes (Helen, Thorsten, Johnathan, Bernard).
These include:
* queries for gene expression (array and in situ) across gross shared anatomical structures in eg. human/mouse/rat and zebrafish/Drosophila developmental time * need for pathway mapping and inference across species * need for annotation source that's multispecies for use in submission tools * need for validated mappings between ontologies that can be used by databases and applications that are not single species - MMIA currently provides these.
- CARO state of the art and experiences extending CARO (David, Thorsten)
CARO has been used successfully, but is not intuitive and could be improved with some small effort. This would promote the use of CARO and ease ontology building for other groups. Specific suggestions are listed in Action Items. A discussion on whether CARO itself should be modified at this point, or whether CARO2 was a preferable route as people are using CARO as a framework to develop other ontologies already. This needs to be discussed further with other interested parties
- Evaluation of CARO, Bilateria and MMIA against presented use cases (Helen, Johnathan, Thorsten, David)
CARO was not intended to solve the problem of multi-species anatomy annotation and representation, nor was it intended to hold maps between ontologies representing anatomy in individual species. The bilateria ontology and the MMIA both address these questions. Both have some relative strengths including the maps to other ontologies present in MMIA (which is not considered to be an ontology and uses only SKOS relationships, broader_than/narrower_than). Clearly building a multi species anatomy ontology is a huge project but work has been done already, there are some low hanging fruit and the use cases are not all anatomically complex.
Action Items
- 1. Generate a list of developmental process terms and submit these to the Gene Ontology for inclusion
E.g. delamination
AI:Johnathan, David, Thorsten
- 2. Development of a common methodology to represent (anatomical) homology groups a David and Thorsten have approached this in a similar way already. In consulatation with phenoscape and others, David has developed a system for representing homology groups using relations based on the "homologous_to" relation defined by phenoscape.
AI:David will generate an OBO file containing his modelling strategy for representing homology groups
AI: David and Thorsten will work together to produce a single system based on their respective modelling approaches. This will include consultations with phenoscape to ensure the new approach is compatible with theirs.
- 3. Development of a Most Recent Common Ancestor Anatomy (MRCA) Ontology
David, Thorsten, Johnathan all have ideas on how to do this, and some work has been done already. It's clear to do significant work there will need to be funding but preliminary work would be very useful and some compact MCRA ontology could be used now by several projects. This is likely a necessary pre-requisite for securing funding. see 6.
- 4. FMA update model.
AI: The conversion of the FMA to OWL format is non trivial and labour intensive. Suzi will contact Onard Medino to discuss this.
- 5. CARO improvements
Thorsten had successfully used CARO to build the bilateria ontology, but it was felt that the use of CARO by other groups would be improved if:
* all the definitions contained examples * an example CARO should be available extended for an example species * a manual on how to use CARO is needed (the paper helps) * the FMA is a good example of definitions plus examples. * synonyms would be useful * epithelial cell could be removed - it's duplicate with the Cell Ontology * CARO doesn't deal with function and therefore organ is not present. David suggested that an abstract class function could be added to help with refining the middle part of CARO * DS and JB will work on defining differentia (labels that would want to attach to a tissue) such as the boundary types that could be included in CARO * communication of any proposed changes should be done by the obo-anatomy list so that all are aware of proposed changes
Post workshop email discussion suggests that CARO is useful, with some more work it could be really useful and easier to use as a framework for development and used as a parent for the emerging cross species anatomy ontology.
AI:David(?) will communicate these suggestions CARO improvements to the CARO team
- 6. Funding
There are two possible sources of funding identified
* an unsolicited R01 * EC funding. This will be conditional on use and implementation, rather than solely development of an ontology. * Datasets that could be integrated include Zebrafish, Gene Expression in situ/array based, VPH
AI: Helen and Bernard will look into EC funding sources (ICAT) AI: Suzi will look into submitting an R01 to develop the MCRA
- 7. MMIA ways forward
MMIA is a useful string matching tool, no ontological claims are made for it.
AI:Johnathan will work with Thorsten, David on a plan for a new CARO derived multi-species anatomy ontology MCRA. See 3.
- 8. Annotation/curator tools
Much of the discussion covered the needs of those using anatomy ontologies rather than those building them. Suggestion that an annotation tool could use some 'lite' version of the MCRA and users could navigate to mapped species specific ontologies if these were needed.
- 9. Visualization
Use of ontologies in developing a visualization methodology/standard e.g. using a schematic body plan to present gene expression and maps to pathways was also discussed. Bernard in particular is interested in this. Stuart Aitken (Edinburgh) has a new grant to develop 2.5D visualization tools. Suzi is interested in generic visualizations that also work for environments.
AI:BdB to contact Stuart Aitken and Richard Baldock about visualization tools.