Limits...
PhenoDigm: analyzing curated annotations to associate animal models with human diseases.

Smedley D, Oellrich A, Köhler S, Ruef B, Sanger Mouse Genetics ProjectWesterfield M, Robinson P, Lewis S, Mungall C - Database (Oxford) (2013)

Bottom Line: As an abundance of phenotype data become available, only systematic analysis will facilitate valid conclusions to be drawn from these data and transferred to human diseases.We show results of an automated evaluation as well as selected manually assessed examples that support the validity of PhenoDigm.Furthermore, we provide guidance on how to browse the data with PhenoDigm's web interface and illustrate its usefulness in supporting research.

View Article: PubMed Central - PubMed

Affiliation: Mouse Informatics Group, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK. ds5@sanger.ac.uk

ABSTRACT
The ultimate goal of studying model organisms is to translate what is learned into useful knowledge about normal human biology and disease to facilitate treatment and early screening for diseases. Recent advances in genomic technologies allow for rapid generation of models with a range of targeted genotypes as well as their characterization by high-throughput phenotyping. As an abundance of phenotype data become available, only systematic analysis will facilitate valid conclusions to be drawn from these data and transferred to human diseases. Owing to the volume of data, automated methods are preferable, allowing for a reliable analysis of the data and providing evidence about possible gene-disease associations. Here, we propose Phenotype comparisons for DIsease Genes and Models (PhenoDigm), as an automated method to provide evidence about gene-disease associations by analysing phenotype information. PhenoDigm integrates data from a variety of model organisms and, at the same time, uses several intermediate scoring methods to identify only strongly data-supported gene candidates for human genetic diseases. We show results of an automated evaluation as well as selected manually assessed examples that support the validity of PhenoDigm. Furthermore, we provide guidance on how to browse the data with PhenoDigm's web interface and illustrate its usefulness in supporting research. Database URL: http://www.sanger.ac.uk/resources/databases/phenodigm

Show MeSH

Related in: MedlinePlus

Determining the phenotype similarity of two entities, e.g. a mouse model and a disease, is a three-step process in our method. The first step is the alignment of ontology concepts based on OWLSim and assigning scores to individual pairs of ontology concepts as illustrated in the top panel of this figure. In a second step, the best scoring matches for each of the annotated ontology concepts are identified and the overall phenotype similarity score described as either the maximum or mean of these scores. In a third step, we scale these two measures relative to their maximum possible values and calculate a single combined percentage score.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3649640&req=5

bat025-F1: Determining the phenotype similarity of two entities, e.g. a mouse model and a disease, is a three-step process in our method. The first step is the alignment of ontology concepts based on OWLSim and assigning scores to individual pairs of ontology concepts as illustrated in the top panel of this figure. In a second step, the best scoring matches for each of the annotated ontology concepts are identified and the overall phenotype similarity score described as either the maximum or mean of these scores. In a third step, we scale these two measures relative to their maximum possible values and calculate a single combined percentage score.

Mentions: Although both human- and mouse-specific data sets were annotated using pre-composed phenotype ontologies (HPO and MP), the zebrafish data were available in a post-composed phenotype description, more specifically the EQ representation (16). To integrate data from all three different data resources, either lexical mappings (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/hp-mp/mp_hp-align-equiv.obo) or the existing logical definitions for HPO (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/hp/hp-equivalence-axioms.obo) and MP (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/mp/mp-equivalence-axioms.obo) were used (9). Five thousand one hundred sixty-two logical definitions for HPO and 6772 logical definitions for MP were used, alongside 2048 pre-calculated lexical matches between HPO and MP. One combined ontology was created to align the individual species-specific annotations with one another. The combined ontology, covering mouse, human and zebrafish data, also included other species-independent OBOFoundry ontologies (http://obofoundry.org/) (30) to bridge between the species-specific ontologies. For instance, UBERON, Neuro-Behaviour Ontology and Phenotypic Quality Ontology were applied in the alignment process (see step one in Figure 1 and section pairwise alignment of ontology concepts with OWLSim). As for the disease and MOD data, all files were downloaded on 1 October 2012. Furthermore, we note here that in a preparation step, the EQ statements used for annotating zebrafish data sets were converted into a pre-composed phenotype presentation [(Zebrafish Phenotype Ontology (ZP)] by assigning an ID and a name to each unique EQ statement.Figure 1.


PhenoDigm: analyzing curated annotations to associate animal models with human diseases.

Smedley D, Oellrich A, Köhler S, Ruef B, Sanger Mouse Genetics ProjectWesterfield M, Robinson P, Lewis S, Mungall C - Database (Oxford) (2013)

Determining the phenotype similarity of two entities, e.g. a mouse model and a disease, is a three-step process in our method. The first step is the alignment of ontology concepts based on OWLSim and assigning scores to individual pairs of ontology concepts as illustrated in the top panel of this figure. In a second step, the best scoring matches for each of the annotated ontology concepts are identified and the overall phenotype similarity score described as either the maximum or mean of these scores. In a third step, we scale these two measures relative to their maximum possible values and calculate a single combined percentage score.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3649640&req=5

bat025-F1: Determining the phenotype similarity of two entities, e.g. a mouse model and a disease, is a three-step process in our method. The first step is the alignment of ontology concepts based on OWLSim and assigning scores to individual pairs of ontology concepts as illustrated in the top panel of this figure. In a second step, the best scoring matches for each of the annotated ontology concepts are identified and the overall phenotype similarity score described as either the maximum or mean of these scores. In a third step, we scale these two measures relative to their maximum possible values and calculate a single combined percentage score.
Mentions: Although both human- and mouse-specific data sets were annotated using pre-composed phenotype ontologies (HPO and MP), the zebrafish data were available in a post-composed phenotype description, more specifically the EQ representation (16). To integrate data from all three different data resources, either lexical mappings (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/hp-mp/mp_hp-align-equiv.obo) or the existing logical definitions for HPO (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/hp/hp-equivalence-axioms.obo) and MP (http://phenotype-ontologies.googlecode.com/svn/trunk/src/ontology/mp/mp-equivalence-axioms.obo) were used (9). Five thousand one hundred sixty-two logical definitions for HPO and 6772 logical definitions for MP were used, alongside 2048 pre-calculated lexical matches between HPO and MP. One combined ontology was created to align the individual species-specific annotations with one another. The combined ontology, covering mouse, human and zebrafish data, also included other species-independent OBOFoundry ontologies (http://obofoundry.org/) (30) to bridge between the species-specific ontologies. For instance, UBERON, Neuro-Behaviour Ontology and Phenotypic Quality Ontology were applied in the alignment process (see step one in Figure 1 and section pairwise alignment of ontology concepts with OWLSim). As for the disease and MOD data, all files were downloaded on 1 October 2012. Furthermore, we note here that in a preparation step, the EQ statements used for annotating zebrafish data sets were converted into a pre-composed phenotype presentation [(Zebrafish Phenotype Ontology (ZP)] by assigning an ID and a name to each unique EQ statement.Figure 1.

Bottom Line: As an abundance of phenotype data become available, only systematic analysis will facilitate valid conclusions to be drawn from these data and transferred to human diseases.We show results of an automated evaluation as well as selected manually assessed examples that support the validity of PhenoDigm.Furthermore, we provide guidance on how to browse the data with PhenoDigm's web interface and illustrate its usefulness in supporting research.

View Article: PubMed Central - PubMed

Affiliation: Mouse Informatics Group, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK. ds5@sanger.ac.uk

ABSTRACT
The ultimate goal of studying model organisms is to translate what is learned into useful knowledge about normal human biology and disease to facilitate treatment and early screening for diseases. Recent advances in genomic technologies allow for rapid generation of models with a range of targeted genotypes as well as their characterization by high-throughput phenotyping. As an abundance of phenotype data become available, only systematic analysis will facilitate valid conclusions to be drawn from these data and transferred to human diseases. Owing to the volume of data, automated methods are preferable, allowing for a reliable analysis of the data and providing evidence about possible gene-disease associations. Here, we propose Phenotype comparisons for DIsease Genes and Models (PhenoDigm), as an automated method to provide evidence about gene-disease associations by analysing phenotype information. PhenoDigm integrates data from a variety of model organisms and, at the same time, uses several intermediate scoring methods to identify only strongly data-supported gene candidates for human genetic diseases. We show results of an automated evaluation as well as selected manually assessed examples that support the validity of PhenoDigm. Furthermore, we provide guidance on how to browse the data with PhenoDigm's web interface and illustrate its usefulness in supporting research. Database URL: http://www.sanger.ac.uk/resources/databases/phenodigm

Show MeSH
Related in: MedlinePlus