Melissa A Haendel; James P Balhoff; Frederic B Bastian; David C Blackburn; Judith A Blake; Yvonne Bradford; Aurelie Comte; Wasila M Dahdul; Thomas A Dececchi; Robert E Druzinsky; Terry F Hayamizu; Nizar Ibrahim; Suzanna E Lewis; Paula M Mabee; Anne Niknejad; Marc Robinson-Rechavi; Paul C Sereno; Christopher J Mungall

“ A major goal for the MODs is to query across anatomical phenotype data to support the identification of relevant genes linked to human diseases (see [10] ) . To facilitate this goal, we constructed the Über Anatomy Ontology (Uberon) , a metazoan multi-species anatomy ontology, to provide a generalization of anatomical structures across the MODs [11] . However, an understanding of biological diversity beyond model organisms necessitates a greater understanding of the variation in morphological form beyond that recorded in the MODs, as well as knowledge about how such variation has evolved. ”

“ Although each taxonomic community could independently add cross-references from their own anatomy ontology to the individual anatomy ontologies of other species, the time and expense alone are prohibitive. Leveraging Uberon as a semantic mapping engine enables immediate access to the rich and expanding genetic and developmental data of the biomedical community that is inherently extensible and computationally accessible. ”

“ Some highly derived organs present in extant taxa are merged in vHOG because they originate from a common ancestral structure. For instance, the swim bladder in ray-finned fish is hypothesized to be homologous to the terrestrial vertebrate lung [37] . While these structures are represented as different terms in Uberon, they are a single term in vHOG, named ‘lung - swim bladder’. Separate terms are maintained in Uberon in such cases, to avoid creating classes representing putative structures or not described in extant species. ”

“ Proposed changes to the core are submitted through the Uberon tracker [39] and vetted by the larger community. Because it is very important to this community and “scientific good practice” to keep track of contributions and to facilitate further discussion regarding design or definitions, a more sophisticated approach to attribution is needed. Contributions to the ontology are marked as metadata in the ontology, at the level of classes, axioms, and the ontology itself, using references to design documents and to individuals using database-cross-reference or ORCID IDs. ”

“ In some cases, the rationale for changing the label was deeper than terminological preference. For example the class UBERON:0006807 was previously labeled ‘lateral epicondyle of humerus’. However, it was pointed out by comparative morphologists on the Phenoscape team that the ‘lateral’ position did not apply to all taxa, due to the changes in posture of limbs that has occurred evolutionarily. We decided to apply a more species-independent terminology and rename this class ‘ectepicondyle of humerus’. ”

“ For every source class that did not have an equivalence axiom in Uberon, we created an obsolescence record, created a new Uberon class, and transferred the axioms across from the source class whilst mapping all entities using the equivalence axioms. ”

“ We therefore documented a set of core design and modeling patterns [32] that would be applied throughout the ontology, and we proceeded to unify the combined ontology along these lines. As a result, all the teeth in Uberon, whether associated with the mandible and maxilla as in mammals or with the vomer as in fishes, have the same relationship (attaches_to) to supporting structures. This is a major advantage for users wishing to query over the combined ontology and provides a modeling pattern that is more stable across taxa. ”

“ In Uberon, joints are defined by the skeletal elements they connect to rather than by the overlap of various skeletal elements, and hence the modeling pattern is more stable because individual bones may vary in their connectivity across species. ”

“ Previously, these vertebrate msAOs contained a mixture of unique and overlapping content. This resulted in duplication of effort, because each defined shared classes such as ‘vertebra’ and ‘nervous system’. In addition, from a user’s perspective, integration was hampered by the need to maintain cross-references or inter-ontology equivalence axioms. At the same time, a number of members of the Phenotype RCN [22] began formalizing expert knowledge in specific areas, such as the neural crest ontology working group [23] and the FEED group working on mammalian feeding muscles [24] . ”

“ To analyze the sequence and nature of evolutionary change across biodiversity, comparisons across basal and derived species within lineages must be made. To study the evolution of development, expression patterns are compared between homologous structures in different species. ”

“ Design documents also have authors and can be included directly in the ontology as instances (see above) via an import file, as well as online and linked via the Uberon.org website.
Discussion
Questions concerning phenotypes are inherently comparative. As examples, to understand the genetic bases of disease, clinical geneticists compare the exomes of people with and without the disease, or biomedical researchers compare the contributions of different genomic elements in a wild-type organism with those of genetically altered ones. ”

“ Because both Uberon and vHOG had their origins in aligning existing species-specific anatomy ontologies, the majority of classes in vHOG were already represented in Uberon. The manually verified groupings in vHOG were transferred into ‘homology notes’ annotations within the Uberon ontology, and were used to refine and verify taxon equivalence axioms. ”

“ Current ontology release files are available always available at: http://purl.obolibrary.org/obo/uberon/releases/
Background
With the rise of the “-omics” age, information, integration, and retrieval have become important challenges for systems biology. Phenomics, or phenotype-based analyses, are a key undertaking, and play an important role in a systems biology approach to understanding gene function and dysfunction. Specifically, we aim to discover the underlying genetic and epigenetic contributions to developmental, evolutionary, behavioral, and morphological variation. ”

“ Another challenge is the use of some terms (such as acropodium) to refer to different structures in different contexts. In some papers “acropodium” only refers to the phalanges (e.g. [29] ) , but in others it refers to the entire manus skeleton excluding the mesopodium (e.g. [30] , [31] ) . In keeping with the definitions created by Haeckel we decided to use ‘acropodium’ (or ‘acropodial skeleton’) for the phalanges and introduce a new term, the ‘digitopodium’, for the metapodium/acropodium complex (Figure 3) . ”

“ In summary, formal integration of anatomical systems from major vertebrate groups into Uberon has expanded and enabled our efforts to investigate disease [46] - [47] and evolution [14] , [18] using comparative anatomical approaches.
Methods Ontology merge
Prior to the merge, we had four source ontologies (VSAO, AAO, TAO, vHOG) and a target ontology (Uberon) . In earlier work, we had enhanced the set of equivalence axioms connecting Uberon classes with source classes. ”

“ We have encountered many difficulties in the process of defining classes of anatomical entities, often because commonly used categories are “fuzzy.” For example, most of the large muscles of the body are “skeletal” muscles. “Skeletal muscles” are attached to skeletal elements, are “voluntary” muscles innervated by somatic motor neurons, and are derived from somites. ”

“ Each class from the set of msAOs is checked for pre-existing equivalence axioms. If there are no equivalence axioms, a new Uberon class is generated; if there is an equivalence axiom, no new class is generated, but the ontology may be augmented with comments if there are differences in structure (e.g. not isomorphic) .Example of class merges. “E” represents class equivalency. (A) Classes and axioms present in target and sources prior to merge. Note that previously Uberon had no classes for the metapterygoid or its joints. (B) Classes present in target post-merge. ”

“ The ontology sources of the classes are indicated, as well as one class that was added post-merge in the new Ext file (‘extracleithrum’) . Note that some classes were merged from multiple sources, such as the class ‘scapula’, which was in all the sources including Uberon, and ‘appendicular skeleton’, which was in VSAO, AAO, and vHOG but not Uberon or TAO. The classes with more sources are shown in increasingly dark gray.Overlap and contributions from source ontologies. A) Venn diagram showing the extent of cross-referenced content between msAOs prior to the merge. ”

“ In contrast, branchiomeric or pharyngeal arch muscles have long been termed “visceral” muscles, because they are part of the gill arch apparatus, which is derived from cranial mesenchyme (cranial paraxial mesoderm) , rather than the somites. The term “visceral” refers to viscera or internal organs. ”

“ Carbon, S; Ireland, A; Bauer, S & Lewis, S (2012) , “Continuous Integration of Open Biological Ontology Libraries”, Bio-Ontologies, ↑ Cui, H (2012) , “CharaParser for fine-grained semantic annotation of organism morphological descriptions”, J Am Soc Inf Sci Technol 5: 738–754, 10.1002/asi.22618, DOI 10.1002/asi.22618 ↑ Image depiction Protégé plugin, ”

“ Brazma, A; et al. (2010) , “Modeling sample variables with an experimental factor ontology”, Bioinformatics 5: 1112–1118, 10.1093/bioinformatics/btq09920200009, pmid:20200009, DOI 10.1093/bioinformatics/btq099 ↑ Cephalopod ontology project, ↑ Chen, C-K; Mungall, CJ; Gkoutos, GV; Doelken, SC; Köhler, S; Ruef, BJ; Smith, C; Westerfield, M; et al. (2012) , “MouseFinder: candidate disease genes from mouse phenotype data”, Hum Mutat 5: 858–866, 10.1002/humu.2205122331800, pmid:22331800, DOI 10.1002/humu.22051 ↑ Smedley, D ”

“ Lapp, H & German, RZ (2011) , “Overview of FEED, the feeding experiments end-user database”, Integr Comp Biol 5: 215–223, 10.1093/icb/icr04721700574, pmid:21700574, DOI 10.1093/icb/icr047 ↑ Shubin, NH; Daeschler, EB & Coates, MI (2004) , “The early evolution of the tetrapod humerus”, Science 5: 90–93, 10.1126/science.109429515064415, pmid:15064415, DOI 10.1126/science.1094295 1 2 3 Hall, BK (2007) , Fins into Limbs: Evolution, Development, and Transformation, Chicago: University of Chicago Press ↑ Haeckel, E (1895) , Systematische Phylogenie. ”