Difference between revisions of "Entities with taxonomic context"

From phenoscape
 
(7 intermediate revisions by the same user not shown)
Line 1: Line 1:
Our phenotype annotations currently employ linking a phenotype composition to a taxon via an ''exhibits'' relation.  However this model has not provided an obvious way to employ taxon-specific homology annotations to the anatomical entities being annotated.  This page describes an alternative representation of phenotype annotations which seems to provide a solution for employing homology annotations as well as anatomical relations which are taxon-specific.
+
Our phenotype annotations currently involve linking a phenotype composition to a taxon via an ''exhibits'' relation.  However this model has not provided an obvious way to employ taxon-specific homology annotations to the anatomical entities being annotated.  This page describes an alternative representation of phenotype annotations which may provide a solution for employing homology annotations as well as anatomical relations which are taxon-specific.
  
 
==Phenotype annotations==
 
==Phenotype annotations==
Line 29: Line 29:
 
==Homology annotations==
 
==Homology annotations==
 
Entity post-compositions provide a method to formulate and employ homology statements:
 
Entity post-compositions provide a method to formulate and employ homology statements:
  intercalarium^in_taxon(Otophysi) homologous_to neural arch 2^in_taxon(Teleostei)
+
  intercalarium^''in_taxon''(Otophysi) ''homologous_to'' neural arch 2^''in_taxon''(Teleostei)
 +
Given the following two phenotype annotations:
 +
intercalarium^''in_taxon''(Danio rerio) ''has_quality'' sigmoid
 +
neural arch 2^''in_taxon''(Chanos chanos) ''has_quality'' serrated
 +
We know that the subjects of these annotations are homologous because these statements are implied:
 +
intercalarium^''in_taxon''(Danio rerio) ''is_a'' intercalarium^''in_taxon''(Otophysi)
 +
neural arch 2^''in_taxon''(Chanos chanos) ''is_a'' neural arch 2^''in_taxon''(Teleostei)
 +
 
 +
==Impact on curation==
 +
Modeling the data in this way shouldn't have an effect on data curators - the interface would be the same in Phenex.  The data would be modeled in this way on its way into OBD.  So, there would be no need for curators to create post-compositions of anatomy terms with taxa; they would just choose the plain anatomy term to use in the phenotype.  These post-compositions are created behind the scenes.  This scenario does open up the possibility of using more relations than just inheres_in/has_phenotype, however.  We could, in the future, make the relationship selectable, allowing curators to input relations between anatomy terms in a more direct manner than using relational qualities.  For example, if there were a relation "fused_with" (not the relational quality), the curator could just make the taxon-specific annotation:
 +
bone_x ''fused_with'' bone_y
 +
 
 +
[[Category:Ontology]]
 +
[[Category:Informatics]]
 +
[[Category:Curation]]
 +
[[Category:Reasoning]]
 +
[[Category:EQ Annotation]]

Latest revision as of 14:26, 30 March 2009

Our phenotype annotations currently involve linking a phenotype composition to a taxon via an exhibits relation. However this model has not provided an obvious way to employ taxon-specific homology annotations to the anatomical entities being annotated. This page describes an alternative representation of phenotype annotations which may provide a solution for employing homology annotations as well as anatomical relations which are taxon-specific.

Phenotype annotations

Old form

Our current phenotype annotation model has the form "a particular taxon has some phenotype, consisting of some quality inhering in some anatomical entity":

Danio rerio exhibits sigmoid^inheres_in(vertebra 1)

The phenotype is a post-composition with the following properties:

sigmoid^inheres_in(vertebra 1) is_a sigmoid
sigmoid^inheres_in(vertebra 1) inheres_in vertebra 1

New form

This annotation can be stated differently as "some anatomical entity, in a particular taxon, has some quality":

vertebra 1^in_taxon(Danio rerio) has_quality sigmoid

The entity is a post-composition with the following properties:

vertebra 1^in_taxon(Danio rerio) is_a vertebra 1
vertebra 1^in_taxon(Danio rerio) in_taxon Danio rerio

Taking into account the contents of the TAO and TTO ontologies, we can infer useful statements such as:

vertebra 1^in_taxon(Danio rerio) is_a vertebra 1^in_taxon(Cyprinidae)
vertebra 1^in_taxon(Danio rerio) is_a bone

Taxonomically variable ontology relationships

Using these entity post-compositions, we can assert taxon-specific anatomical relationships such as:

vertebra 1^in_taxon(Cyprinidae) part_of Weberian apparatus

Now if we search for phenotype annotations in which the subject entity is part_of the Weberian apparatus, we will find the one about vertebra 1 of Danio rerio:

vertebra 1^in_taxon(Danio rerio) has_quality sigmoid

However, we would correctly not find an annotation about vertebra 1 in Chanos chanos, which is not in Cyprinidae:

vertebra 1^in_taxon(Chanos chanos) has_quality serrated

Searching for phenotype annotations in which the subject entity is vertebra 1 would return both statements.

Homology annotations

Entity post-compositions provide a method to formulate and employ homology statements:

intercalarium^in_taxon(Otophysi) homologous_to neural arch 2^in_taxon(Teleostei)

Given the following two phenotype annotations:

intercalarium^in_taxon(Danio rerio) has_quality sigmoid
neural arch 2^in_taxon(Chanos chanos) has_quality serrated

We know that the subjects of these annotations are homologous because these statements are implied:

intercalarium^in_taxon(Danio rerio) is_a intercalarium^in_taxon(Otophysi)
neural arch 2^in_taxon(Chanos chanos) is_a neural arch 2^in_taxon(Teleostei)

Impact on curation

Modeling the data in this way shouldn't have an effect on data curators - the interface would be the same in Phenex. The data would be modeled in this way on its way into OBD. So, there would be no need for curators to create post-compositions of anatomy terms with taxa; they would just choose the plain anatomy term to use in the phenotype. These post-compositions are created behind the scenes. This scenario does open up the possibility of using more relations than just inheres_in/has_phenotype, however. We could, in the future, make the relationship selectable, allowing curators to input relations between anatomy terms in a more direct manner than using relational qualities. For example, if there were a relation "fused_with" (not the relational quality), the curator could just make the taxon-specific annotation:

bone_x fused_with bone_y