94 5 Survey of Ontologies in Bioinformatics
Because GO supports two relationships, its ontologies are more expressive
than a taxonomy. However, modern ontologies often support many more
than just two relationships. An analysis of the names of GO terms suggests
that there are many other relationships implicitly contained in the GO ter-
minology. For example, 65.3% of all GO terms contain another GO term as
a proper substring. This substring relation often coincides with a deriva-
tional relationship between the terms (Ogren et al. 2004). For example, the
termregulation of cell proliferation(GO:0042127) is derived from the termcell
proliferation(GO:0008283) by addition of the phraseregulation of. The phrase
regulation ofoccurs frequently in GO, yet is not itself a GO term. Furthermore,
this subterm occurs consistently in different subsets of the GO ontologies.
Derivational subterms such as this one indicate interesting semantic rela-
tionships between the related terms. Formalizing these relationships would
result in a richer representation of the concepts encoded in the ontology, and
would assist in the analysis of natural language texts.
Many programs have been developed for profiling gene expression based
on GO or the GO file format. These programs have been very useful for
translating sets of differentially regulated genes.
DAG-Edit
sourceforge.net/project/showfiles.php?group_id=36855
DAG-Edit is an open source tool written in Java for browsing, searching,
and modifying structured controlled vocabularies. DAG-Edit was previ-
ously called GO-Edit. It is applicable to any kind of structured controlled
vocabulary. Three formats are supported: GO flat file, GO serial file, and
OBO file.
GenMAPP http://www.GenMAPP.org
This tool visualizes gene expression and other genomic data on maps repre-
senting biological pathways and groupings of genes. Integrated with Gen-
MAPP are programs to perform a global analysis of gene expression or ge-
nomic data in the context of hundreds of pathway MAPPs and thousands
of GO Terms (MAPPFinder), import lists of genes/proteins to build new
MAPPs (MAPPBuilder), and export archives of MAPPs as well as expression
and genomic data to the web.
GoMiner discover.nci.nih.gov/gominer
This program package organizes lists of “interesting” genes (e.g., under- and
overexpressed genes from a microarray experiment) for biological interpre-
tation (Zeeberg et al. 2003). GoMiner provides quantitative and statistical
output files and two useful visualizations. The first is a treelike structure