Biology Reference
In-Depth Information
rators” or “editors” who have a broad knowledge of the overall struc-
ture, scope, completeness, and consistency requirements of the ontology
and maintain ultimate control over its terms.
23
The GO has certainly not solved all communication and consistency
problems in biology. This is, in part, because it was designed only to pro-
vide a language for talking about gene function. The success of the GO
has inspired a host of ontologies in other biological domains: cell types,
descriptions of environment, experimental techniques, human diseases,
anatomy, pharmacogenomics, imaging methods, pathways, and human
phenotypes all have their own ontologies. The Open Biomedical On-
tologies (OBO) Foundry attempts to bring order to this proliferation
by setting up rules and standards for the creation of ontologies them-
selves, with the ultimate aim being “a suite of orthogonal interoperable
reference ontologies in the biomedical domain.”
24
In particular, OBO
has developed a “relationship types” ontology that specifi es the logi-
cal connectors (“is_a,” “part_of,” “has_part,” “has_agent,” etc.) used by
ontologies.
25
Such ontologies of ontologies, or meta-ontologies, provide
a framework for what their creators hope will be an all-encompassing
language for describing biology and medicine.
Proponents of ontologies believe that they are the only reliable way
of promoting the free exchange of experimental data in science. To make
this point, Barry Smith (one of the founders of the National Center for
Biomedical Ontologies) contrasts ontologies with “folksonomies” of the
kind enabled by the photo-sharing website Flickr.
26
Photos in Flickr can
be “collaboratively categorized” using “freely chosen keywords” based
on the tags that individuals apply to their photos.
27
How is it possible to
choose among the myriad of individual sets of classifi cations if there are
no constraints? Smith examines four possibilities: deferring to the au-
thority of a “terminology czar,” using the fi rst that comes along, using
the best, or using an ontology based on “reality, as revealed incremen-
tally by experimental science.”
28
An “instrumental” ontology, based on
a particular model of how biology works, will sooner or later be super-
seded or outcompeted. A “realist ontology,” Smith argues, is the only
way to make ontologies work. Terms in an ontology should correspond
to “what one actually sees in a lab, not what is convenient”
29
:
Ontology, as conceived from the realist perspective, is not a soft-
ware implementation or a controlled vocabulary. Rather, it is a
theory of reality, a “science of what is, of the kinds and struc-
tures of objects, properties, events, processes and relations in
every area of reality.”
30
