Using Biomedical Ontologies for Data Representation and Man-
        agement in the Mouse Genome Informatics (MGI) System
              Li Ni,* Carol J. Bult, Jim A. Kadin, Joel E. Richardson, Martin Ringwald, Janan T. Eppig, Judith A. Blake
                                          The Jackson Laboratory, Bar Harbor, Maine, U.S.A.



ABSTRACT                                                            allows consistent annotation of mouse genotypes with
                                                                    standard phenotype MGI is also the originator of the Adult
Structured vocabularies and ontologies are increasingly uti-
                                                                    Mouse Anatomy (MA), it allows us to navigate through the
lized for managing, annotating, and analyzing complex bio-
                                                                    extensive dictionary hierarchies for the different develop-
logical data. The Mouse Genome Informatics (MGI,
                                                                    mental stages, to locate specific anatomical structures within
http://www.informatics.jax.org) resource contributes to and
                                                                    those hierarchies, and to annotate an obtain the expression
utilizes a variety of vocabularies and ontologies to robustly
                                                                    results associated with those structures.
capture and provide biological information about the labora-
tory mouse and its use as a model for human disease. Here
                                                                    MGI also incorporates Online Mendelian Inheritance in
we report on the power and complexity of using biomedical
                                                                    Man (OMIM) terms to make similarity assertions between
ontologies for data representation and management at MGI.
                                                                    mouse models and human disease; incorporates the Mouse
MGI is a highly integrated database and software system
                                                                    Anatomical Dictionary for describing expression data dur-
integrating mouse genetic, genomic and phenotypic infor-
                                                                    ing mouse embryonic development.
mation including data on gene characterization, pathways,
protein classifications, sequence, gene expression, alleles,
                                                                    Additionally, for mouse gene products functional annota-
phenotypes, diseases, and comparative gene data for mouse,
                                                                    tion, MGI start to use the Cell Ontology, the Mouse Embry-
human, rat and other mammals.
                                                                    onic and Adult Anatomy Ontologies, the Evidence Code
                                                                    Ontology, and PSI-mod to add additional information to a
The Gene Ontology (GO, http://www.geneontology.org) is
                                                                    Gene Ontology annotation. These combined ontology anno-
the most widely used vocabulary for providing connections
                                                                    tations will be loaded into the database in the near future.
between proteins and their roles in the biological organiza-
tion. MGI is a founding member of the GO Consortium, and
                                                                    The structured vocabulary-based annotations assist in robust
actively participates in the ongoing development of GO, as
                                                                    and accurate data mining when posing such complex ques-
well as applying GO to functional annotation of mouse gene
                                                                    tions in both computational and individual formats at MGI.
products. MGI provide an automatically generated text re-
                                                                    Though the use of multiple ontologies, MGI is able to ro-
port, a tabular form view and a graphical display of GO an-
                                                                    bustly represent many components of knowledge about the
notations in the gene function detail page. MGI also provide
                                                                    mouse model system. While these ontologies are inde-
comparative graphs of GO annotations for mammalian
                                                                    pendently developed, the concurrent use of them within the
orthologs. In addition, MGI hosts graphical and statistical
                                                                    MGI system illuminates some challenges in the intersection
tools exploiting the hierarchical structure of the GO to aid in
                                                                    of ontologies that MGI ontology developers and curators
visualization of annotations and the analysis of large data
                                                                    work to address. For example, the MP and the GO termi-
sets such as microarray data.
                                                                    nologies incorporate mouse anatomical terms from the
                                                                    MA. Testing and updating the MP and the GO to accurately
MGI has spearheaded the development of two other major
                                                                    reflect the canonical anatomy organization in the MA re-
structured vocabularies in support of annotation and analysis
                                                                    quires resources and attention on a regular basis.
of mouse biology: the Mammalian Phenotype (MP) Ontolo-
gy, a widely adopted ontological model that enables pheno-
                                                                    MGI is a consortium of database resources with funding
type annotations to background-specific allelic genotypes at
                                                                    obtained from NHGRI (HG00330 and HG02273),
varying degrees of granularity, this structured vocabulary
                                                                    NIH/NICHD (HD062499), and NCI (CA89713).
*
    To whom correspondence should be addressed: li.ni@jax.org



                                                                                                                                1