Next: Building an Ontology
Up: A Survey of Current
Previous: The TAMBIS Ontology
There are two important messages from this brief survey of bio-ontologies: the first is that
ontologies are being used within the community to provide knowledge input to
databases and applications. The second message is that all these ontologies are
very different and specific to their intended use. TaO is an ontology of bioinformatics tasks
and so contains such concepts as AccessionNumber and
ProteinId, which are not part of the world of molecular biology.
The TaO could not be substituted for EcoCyc's ontology.
GO is an ontology of gene product function and
RiboWeb represents knowledge of Ribosomal subunit structure, data and
methodologies. As GO is used for database annotation, it holds a fine level of
detail whereas the TaO is quite shallow, but precision is gained during query
formulation by joining concepts together.
Even if one ontology could be developed, individual applications would only use a
subset, leading to a requirement of highly modular ontologies with minimised
dependencies and assumptions between them. That ontology use
influences the content and nature of the knowledge captured within an
ontology is not a contradiction of the knowledge holding ability of
ontologies. Not only does the purpose determine the scope and
granularity to which the same knowledge is represented in different
ontologies, but conceptualisations may differ without one being
incorrect. For example, TaO describes that DNA may be translated to
protein. This is wrong in molecular biological terms, but is a feature
of bioinformatics - so conceptualisations of the same domain may
differ. Sometimes a constraint is necessary for an application and
sometimes it is not needed for another, this simply changes what
knowledge is captured or how it is captured, it does not change the
knowledge itself.
Next: Building an Ontology
Up: A Survey of Current
Previous: The TAMBIS Ontology
Robert Stevens
2001-07-19