The Mammalian Phenotype Ontology as a tool for annotating, analyzing and comparing phenotypic information
© Smith et al.; licensee BioMed Central Ltd. 2004
Received: 8 September 2004
Accepted: 17 November 2004
Published: 15 December 2004
The Mammalian Phenotype (MP) Ontology enables robust annotation of mammalian phenotypes in the context of mutations, quantitative trait loci and strains that are used as models of human biology and disease. The MP Ontology supports different levels and richness of phenotypic knowledge and flexible annotations to individual genotypes. It continues to develop dynamically via collaborative input from research groups, mutagenesis consortia, and biological domain experts. The MP Ontology is currently used by the Mouse Genome Database and Rat Genome Database to represent phenotypic data.
Mammalian phenotypes are complex and the term itself is imprecise. Generally, we use the word phenotype in referring to the appearance or manifestation of a set of traits in an individual that result from the combined action and interaction of genotype and environment.
Because mouse is the premier model organism for the study of human biology and disease, the goal of comparative phenotyping and building new animal models through genetic engineering holds great promise. The mouse has distinct advantages for studies that translate to humans. It is a small, short-lived mammal with a fully sequenced genome in which all life stages can be accessed, and for which myriad tools are available for precisely experimentally manipulating its genome. Further, the large collection of inbred strains of mice and the controlled environment in which the animals live provides the ability to confirm phenotype observations and to systematically perturb environmental factors and genetic input to measure effects under defined conditions. Current international efforts to 'make a mutation' for every gene through mutagenesis  and genetic engineering [2, 3] make it imperative for phenotype data to be represented in standard descriptive formats to enable computational analysis and comparison.
Mammalian phenotypes are frequently genetically complex. Mutation of even a single gene almost always produces pleiotropic effects. Conversely, non-allelic mutations can produce indistinguishable phenotypes. Modifier genes and epistatic interactions can markedly alter the phenotype. Combining different allelic combinations of different genes can produce unique phenotypes not found in the single-gene mutation genotype. Imprinting of genes can dramatically affect phenotype. Mutations expressed in different inbred strains of mice can manifest as an increase or decrease of severity or penetrance of the corresponding phenotype. Quantitative trait loci (QTL) can contribute in complex nonlinear ways to the phenotype. In addition, mutations that are 'genomic' in nature, either disrupting or deleting multiple genes or occurring in intergenic regions, can produce distinct phenotypes and challenge us to think beyond gene effects to genomic effects. The outcome of these complex interactions can be dissected and reproducibly examined by characterizing inbred strains that represent the combined phenotype of the 'whole-genome' genotype in its environmental context.
The Mouse Genome Database (MGD) at the Mouse Genome Informatics website [4, 5] serves as the model organism database for mouse, representing the genetics, genomics and biology of the mouse and as a community resource for mammalian studies. Significant reorganization and modeling of phenotypes is now underway to support these data robustly, to represent phenotypes in ways that are computationally accessible, and to provide human interfaces to these data that will enable knowledge building and hypothesis generation. One component of this work is the development of the Mammalian Phenotype (MP) Ontology, a structured vocabulary that will aid in standardizing annotations and, with its concepts definitions, unambiguously describe phenotypic observations.
Results and discussion
The problems of text
Written descriptions of phenotypes in higher organisms reflect the complexity of the subject, the richness of language, and the phenomenal diversity that these data represent. While text descriptions are commonly used in publications describing phenotype, and have been the basis of electronically accessible phenotypic descriptions (for example, Online Mendelian Inheritance in Man (OMIM)  and the Mouse Locus Catalog (MLC) , text is unreliable for searching, either manually or computationally. From the user's perspective, even the best full-text search including Boolean operators will miss appropriate records (false negatives) and return unwanted records (false positives).
Text search for mutations with hair/fur loss defects in the Mouse Locus Catalog
Loss of hair
Loss of fur
Composite 'true' result
27 true results
A further detriment to database text records is their difficulty to update and maintain. As new information is learned about a phenotypic mutant, the record must be continually rewritten. Although this practice might be sustained for a small number of records, it does not scale when thousands of mutant records are considered. The alternative of simply adding on another paragraph to existing text records becomes confusing, with potentially conflicting information and different writing styles appearing in one textual description, and unwieldy, with more and more text that may no longer represent a logical synthesis.
Nomenclatures, vocabularies and ontologies
Formal nomenclatures for genes, mutant alleles and inbred strains of mice have existed since the 1940s [8, 9]. The MGD  serves as the authoritative source for the names and symbols associated with mouse genes, alleles and strains. The advantage of applying such nomenclatures has been increasingly recognized as genomes become better defined and the realized power of comparative genomics allows homologous and orthologous gene relationships to be explicitly defined. At present, human, mouse and rat gene nomenclatures operate in parallel, using coordinated symbols for all three species' genes. In addition, mouse and rat strain nomenclatures were merged to one standard strain nomenclature recently, making strain identity and nomenclature conventions consistent. Nomenclature guides for mouse and rat genes, mutant alleles, and strains are available online and regularly revised based on international nomenclature committees' reviews .
Beyond nomenclatures, which are key to object identities and relationships, are vocabularies that can be used to describe broader concepts and categorizations. Vocabularies can take many forms, including simple lists of controlled terms, such as the cytogenetic band designations used to name the bands defined by chromosome staining or the classes of genetic markers, such as gene, pseudogene, expressed sequence tag (EST), and so forth.
The annotation of complex biological data and concepts requires more than lists and simple vocabularies. Ontologies, or 'descriptions of what there is', contain both concepts, with precise meanings, and relationships among those concepts. As such ontologies are able to support descriptions of complex biology and are useful in making these data more amenable to computational analyses. The first widely used ontology developed and adopted in the biological domain is the Gene Ontology (GO) [11–13] which contains concepts of molecular function, cellular localization and biological process for annotating the functional aspects of genes. The GO is structured as a directed acyclic graph (DAG), where each vocabulary term (node) may have both multiple parent term and multiple child term relationships. MGD uses GO extensively for gene annotation . In addition, MGD has adopted the Mouse Embryo Anatomy Nomenclature Database  and the Anatomical Dictionary for the Adult Mouse  for annotating data that include anatomical attributes, such as tissue sources for clones and phenotypes. The Gene Expression Database (GXD) , integrated with MGD through the Mouse Genome Informatics (MGI) system , applies these anatomical ontologies as a central concept in the description of expression data.
Mammalian Phenotype Ontology
Although the need for vocabularies as key components to consistent phenotype annotations for mammals has been recognized for some time , and many smaller controlled vocabularies have been implemented to describe various aspects of phenotype in MGD (for example, class of mutation, embryonic stem (ES) cell lines used for generating targeted mutations, type of inheritance), much of the data has remained in text form. Over the past two years, the Mammalian Phenotype (MP) Ontology has emerged to more precisely describe phenotypes, and to allow easier access to phenotype-sequence interactions.
Our goal is to describe the richness of phenotypes as precisely as they are known, recognizing that phenotype data are by nature complex and usually incomplete. Taking advantage of structural properties of a DAG, we have the ability to annotate phenotypes to the level of data resolution available, whether general or very specific and the ability to query with a high-level term, returning all phenotypes containing annotations to that term or to terms more specific than the query term. Thus, one can query for 'respiratory signs/symptoms' and retrieve all phenotypes annotated to this term and its hierarchical 'children' (abnormal breathing, abnormal respiratory sounds, anoxia, apnea, dyspnea, hypercapnia, and so on), or specifically request annotations to any of these sub-terms.
Each MP ontology term has a unique identifier, a definition and synonyms. In the term detail pages, these data and the number of hierarchical paths of the vocabulary where the terms appear are displayed. A plus sign following the term indicates that children of this term exist. In this figure, displayed next to the term, is a link indicating the number of annotation instances in MGD using this term or children of this term. This feature, due to be publicly available in early 2005, will greatly improve phenotype-centric searching in MGD.
Developing the MP vocabulary
To initiate the vocabulary, we first developed a high-level categorization of phenotypes consisting of approximately 100 terms, such as heart/cardiovascular dysmorphology and skeletal axial defects. As we used this list for annotations, terms were refined and general organizing principles for the MP vocabulary were developed.
An important component of our approach has been to address two practical implementation questions. From the biologist's perspective, the question is what term would be used to describe a specific phenotypic trait. From the curation perspective, we ask what terms reflect biological reality and maximize curator productivity.
MP Ontology terms and their decomposed forms
MP Ontology term*
Body location (anatomy/cells)
Brain cerebral ventricles
Dystrophic cardiac calcinosis‡
Eye lens capsule
Three major strategies are being pursued to further develop the vocabulary itself. First and most important is through the ongoing process of curating phenotype data. As new phenotypic traits are described and published, the need for new terms is recognized. New terms added in this way may be a simple addition to an existing hierarchical path or may result in the addition of entire new branches in the hierarchy. Second, collaborative efforts between the MGD phenotype curators, the mouse mutagenesis centers and the rat genetics community identify new specific terms and suggest improved organization of terms within particular hierarchical branches. Third, we are recruiting individuals with expertise in specific biological domains to review and evaluate sections of the vocabulary for accuracy, completeness and systematic arrangement. The MP Ontology is a work in progress and remains incomplete in some areas. We welcome the participation of the mammalian research community so that the most useful, definitive and universally applicable terms will be included. Information can be obtained by sending e-mail to email@example.com.
While common pathological and clinical terms are used in the MP Ontology, considerations for term placement within the structure and for precise terminology is often derived from comparison with other open biological ontologies (OBO) . Recently, a cell-type ontology has become available  and a comparison of terminology to this ontology has not yet been completed. We are working with the mutant mouse pathology database Pathbase [24, 25] to map and cross-reference terms from their Pathology Ontology.
The MP Ontology was built as a DAG using the DAG-Edit software written by John Richter and Suzanna Lewis . The MP Ontology is updated daily and can be browsed or searched online at . MP files also are available in flat file format and OBO format from our ftp site  and are posted at the OBO site .
Phenotype data annotation
The MP Ontology and annotation schema was designed to minimize curatorial time, yet remain precise enough to describe phenotypic data. It supports robust phenotypic annotations and querying capabilities for mouse phenotype data. While this vocabulary is far from complete, we have designed strategies for its continued development as a collaborative effort for supporting the representation of existing mutations and those that continue to be created.
As of 1 November 2004, over 11,150 phenotypic alleles representing mutations in 5,214 unique genes had been catalogued in MGD. For these alleles, 9,696 genotype records exist, with 21,556 phenotypic annotation instances. The MP Ontology is also used in phenotypic data annotations at the RGD . As our database groups continue to accumulate annotations, it will be possible to mine these data to ask interesting questions about similarities and differences in comparable allele effects between the species, as well as within species. Comparative phenotype data will potentially uncover new modifier effects and point to new pathway relationships and genetic networks tied to disease processes. The MP Ontology will be critical for enabling computational analyses and providing a framework for improved web views and other human-comprehensible displays for the research community.
We thank Martin Ringwald and Susan Bello for helpful comments on the manuscript, and acknowledge members of the MGI and RGD teams, Neuromice consortium and Richard S. Smith for contributions to the MP Ontology development.
- Nadeau JH, Balling R, Barsh G, Beier D, Brown SD, Bucan M, Camper S, Carlson G, Copeland N, Eppig J, et al: Sequence interpretation. Functional annotation of mouse genome sequences. Science. 2001, 291: 1251-1255. 10.1126/science.1058244.View ArticleGoogle Scholar
- Austin CP, Battey JF, Bradley A, Bucan M, Capecchi M, Collins FS, Dove WF, Duyk G, Dymecki S, Eppig JT, et al: The knockout mouse project. Nat Genet. 2004, 36: 921-924. 10.1038/ng0904-921.View ArticleGoogle Scholar
- Auwerx J, Avner P, Baldock R, Ballabio A, Balling R, Barbacid M, Berns A, Bradley A, Brown S, Carmeliet P, et al: The European dimension for the mouse genome mutagenesis program. Nat Genet. 2004, 36: 925-927. 10.1038/ng0904-925.View ArticleGoogle Scholar
- MGI_3.01 - Mouse Genome Informatics. [http://www.informatics.jax.org]
- Bult CJ, Blake JA, Richardson JE, Kadin JA, Eppig JT, Baldarelli RM, Barsanti K, Baya M, Beal JS, Boddy WJ, et al: The Mouse Genome Database (MGD): integrating biology with the genome. Nucleic Acids Res. 2004, 32 (Database issue): D476-D481. 10.1093/nar/gkh125.View ArticleGoogle Scholar
- Online Mendelian Inheritance in Man (OMIM). [http://0-www.ncbi.nlm.nih.gov.brum.beds.ac.uk/entrez/query.fcgi?db=OMIM]
- Mouse Locus Catalog (MLC). [http://www.informatics.jax.org/searches/noforms_mlc_omim.cgi]
- Dunn LC, Gruneberg H, Snell GD: Report of the committee on mouse genetics nomenclature. J Hered. 1940, 31: 505-506.Google Scholar
- Snell GD: Biology of the Laboratory Mouse. 1941, New York: McGraw-Hill, 1Google Scholar
- Mouse Nomenclature Home page. [http://www.informatics.jax.org/nomen]
- Gene Ontology Consortium. [http://www.geneontology.org]
- Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, et al: Gene Ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000, 25: 25-29. 10.1038/75556.View ArticleGoogle Scholar
- Gene Ontology Consortium: Creating the gene ontology resource: design and implementation. Genome Res. 2001, 11: 1425-1433. 10.1101/gr.180801.View ArticleGoogle Scholar
- Hill DP, Davis AP, Richardson JE, Corradi JP, Ringwald M, Eppig JT, Blake JA: Strategies for biological annotation of mammalian systems: implementing gene ontologies in mouse genome informatics. Genomics. 2001, 74: 121-128. 10.1006/geno.2001.6513.View ArticleGoogle Scholar
- Bard JL, Kaufman MH, Dubreuil C, Brune RM, Burger A, Baldock RA, Davidson DR: An internet-accessible database of mouse developmental anatomy based on a systematic nomenclature. Mech Dev. 1998, 74: 111-120. 10.1016/S0925-4773(98)00069-0.View ArticleGoogle Scholar
- Hayamizu TF, Magan M, Corradi JP, Kadin JA, Ringwald M: The Anatomical Dictionary for the Adult Mouse: a tool for annotating and integrating data. Genome Biol.Google Scholar
- Hill DP, Begley DA, Finger JH, Hayamizu TF, McCright IJ, Smith CM, Beal JS, Corbani LE, Blake JA, Eppig JT, et al: The mouse Gene Expression Database (GXD): updates and enhancements. Nucleic Acids Res. 2004, 32 (Database issue): D568-D571. 10.1093/nar/gkh069.View ArticleGoogle Scholar
- Eppig JT: Algorithms for mutant sorting: the need for phenotype vocabularies. Mamm Genome. 2000, 11: 584-589. 10.1007/s003350010111.View ArticleGoogle Scholar
- Mammalian Phenotype Ontology Search. [http://www.informatics.jax.org/searches/MP_form.shtml]
- Gkoutos GV, Green ECJ, Hancock JM, Davidson D: Using ontologies to describe mouse phenotypes. Genome Biol. 2004, 6: R8-10.1186/gb-2004-6-1-r8.View ArticleGoogle Scholar
- Gene Ontology Consortium: The Gene Ontology (GO) database and informatics resource. Nucleic Acids Res. 2004, 32 (database): D258-D261. 10.1093/nar/gkh036.View ArticleGoogle Scholar
- Open biological ontologies (OBO). [http://obo.sourceforge.net]
- Bard JL, Ashburner M: A Structured Controlled Vocabulary of Cell Types. http://obo.sourceforge.net/cgi-bin/detail.cgi?celltype,
- Pathbase. [http://www.pathbase.net]
- Schofield PN, Bard JB, Booth C, Boniver J, Covelli V, Delvenne P, Ellender M, Engstrom W, Goessner W, Gruenberger M, et al: Pathbase: a database of mutant mouse pathology. Nucleic Acids Res. 2004, 32 (Database issue): D512-D515. 10.1093/nar/gkh124.View ArticleGoogle Scholar
- DAG-Edit. [http://www.geneontology.org/doc/GO.tools.html#dagedit]
- Mammalian Phenotype Ontology ftp site. [ftp://ftp.informatics.jax.org/pub/reports/index.html#pheno]
- Mammalian Phenotype Ontology posted at the OBO site. [http://obo.sourceforge.net/cgi-bin/detail.cgi?musphen]
- Rat Genome Database (RGD). [http://rgd.mcw.edu]
- MGI record for the Ccm1 tm1Dmar phenotypic allele. [http://www.informatics.jax.org/searches/allele.cgi?22569]
- MGI Alleles and Phenotypes Query Form. [http://www.informatics.jax.org/searches/allele_form.shtml]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.