Special Report 19

Ontology development

Formalising the associative relationships between an anatomical structure and its spatial location, functional system and chronological stage requires hierarchical information. Ontologies link facts as a triad of related terms that can be integrated with other data using common controlled vocabularies (Smith et al., 2007). This can be done using web-accessible resources such as CARO (Common Anatomy Reference Ontology), CL (Cell Type), ZFA (Zebrafish Anatomy and Development), and EMAP (Mouse Gross Anatomy and Development), which can be found at The OBO and OWL hot-links found at http://obofoundry.org/.

Building a formal system that unambiguously makes explicit the knowledge to be included in the ontology of developmental processes and toxicities is not a trivial task (Bard, 2005). To bring together the vertical observational series (e.g. phenotype ontology) with a longitudinal embryological series (e.g. the forward progression of outcomes as development advances) is a composite task. For example, existing ontologies can be merged and thus arrange information by embryology (EMAP) and developmental toxicology (DevTox). Thus ToxRefDB taxonomises 982 terms (level-5) into 51 embryological targets (level-4), 24 embryological systems (level-3), 141 tissue localisations (level-2), and 3 observational modes (level-1 modes). The model combined DevTox ontology (levels-1, -2 and -5) with developmental ontology from EMAP (level-3, embryological system; level-4, embryological target). The Open Biomedical Ontology (OBO) website[1] in Web Ontology Language (OWL)[2] format has also been used to write developmental ontologies for Theiler Stages (see Bard, 2007) and Carnegie Stages (see Hunter et al., 2003), describing mouse and human development, respectively.

Having a sound DevTox ontology will codify the organisation of facts and concepts into useful descriptions based on embryology and some degree of common pathogenesis and interoperability with other resources. For example, an emerging mouse/mammalian phenotype ontology resource using OBO is being developed for the Mammalian Phenotype Ontology (MPO) browser as part of the Mouse Genome Informatics (MGI) project at The Jackson Laboratory (http://www.informatics.jax.org/). The MPO browser deconstructs mammalian phenotypes into their constituent terms using a schema proposed as phenotype/attribute (PATO[3], EUMORPHIA[4])/value, as shown in the example below.

MGI EXAMPLE: microphthalmia (199 genotypes, 199 annotations)

<MP term>                             microphthalmia

<synonym>                            small eyes

<MP id>                                    MP:0001297

<Definition>        reduced average size of the eyes

The same condition is represented in OBO as:

[Term]

id: MP:0001297 !              microphthalmia

intersection_of:                 PATO:0000587 ! decreased size

intersection_of:                 inheres_in MA:0000261 ! eye

A main advantage of using PATO is the ability to express phenotypic ontologies based on concept relationships, rather than instances. A PATO-compliant zebrafish database is being developed by The Jackson Laboratory to manage morpholino-induced phenotypes (morphants)[5] (Knowlton et al., 2008).

  • Informal ontologies

[1] OBO website: http://obofoundry.org/

[2] OWL is a language of the semantic web to express natural language (used on the world-wide web) in machine-readable form. It uses a triad structure to define classes and interrelationships to annotate taxonomic hierarchy <classes><properties><individuals>:

                                             <classes>            unit of taxonomy; sets, collections, or types of objects

                                             <properties>      features that objects have and share (attributes)

                                             <individuals> basic, ground level objects (entities); instances in a class

[3] PATO: Phenotypic Attribute Trait Ontology.

[4] EUMORPHIA: A project that connects mouse mutants to human disease.

[5] A type of molecule used in molecular biology that alters the development of genes by preventing access by other molecules (knockdown of targeted gene expression).

[7] PATO: Phenotypic Attribute Trait Ontology.

[8] EUMORPHIA: A project that connects mouse mutants to human disease.

[9] A type of molecule used in molecular biology that alters the development of genes by preventing access by other molecules (knockdown of targeted gene expression).