Table 1.

General overview of the method

Step	Method	Materials used	Example
Formalization of cross-species anatomy	Assign equivalent class statements based on UBERON cross-species mappings; integrate all ontologies and equivalence statements in single ontology.	Gene Ontology, Mouse Anatomy, Worm Anatomy and Development, Zebrafish Anatomy and Development, Fly Anatomy, Foundational Model of Anatomy, UBERON	Both the classes ‘Tail’ from Mouse Anatomy and ‘Caudal fin’ from Zebrafish Anatomy are declared equivalent to the class ‘Tail’ in UBERON.
Consistency verification and removal of contradictions	Remove disjointness statements from UBERON.	UBERON ontology, processing of OBO Flatfile Format	The disjointness between ‘Material anatomical entity’ (UBERON:0000465) and ‘Immaterial anatomical entity’ (UBERON:0000466) is removed.
Formalization of cross-species phenotypes	Convert phenotype ontologies' definitions to enable interoperability with anatomy (using has-part and part-of relations); combine all phenotype ontologies, their class definitions and the cross-species anatomy ontology in a single ontology.	Yeast Phenotype, FlyBase Controlled Vocabulary, Worm Phenotype, Mammalian Phenotype, Human Phenotype; related ontologies: PATO, ChEBI, Gene Ontology, Mouse Pathology, Celltype, Protein Ontology	Define the mouse phenotype ‘Matted coat’ as shown in Figure 2.
Represent phenotype annotations	Phenotype annotations in model organisms databases or of diseases are represented as class C; the class C is then asserted as equivalent to the intersection of the annotated phenotypes.	HPO-based phenotype annotations of OMIM; phenotype annotations in model organism databases	The class ‘Alport syndrome’ is defined as equivalent to the intersection of the disease's phenotypic characteristics: ‘Renal failure, Nephritis, Hearing loss’ and ‘Hematuria’.
Inference of cross-species phenotype representation	Using automated reasoning, each class that represents a genotype annotations or disease is examined and all its super-classes in each species-specific phenotype ontology are inferred. The result is a representation of the annotated phenotypes based on five species-specific phenotype ontologies.	OWL reasoners (CB and CEL)	The worm phenotype ‘Abnormal apoptosis’ is inferred as a super-class of the human phenotype ‘Defective lymphocyte apoptosis’.
Application of semantic similarity	A semantic similarity measure is applied to compensate for missing information and noisy data.	Jaccard metric weighted by information content; implemented parallel algorithm for computation	The phenotype of an allele of the Adam19 gene (MGI:3028702) is similar to tetralogy of Fallot phenotype.
Quantitative evaluation	KEGG and known disease models provide gene–gene and gene–disease associations which are compared within the network. Orthologous genes and genes in the same pathway are phenotypically similar; gene–disease pairs of known gene–disease associations are similar.	KEGG, OMIM and Morbidmap, mouse model annotations in MGI	The area under the receiver operator characteristic curve (a plot of the true positive rate as function of the false positive rate) for pathways is 0.59, for orthology 0.62 and for disease 0.68.