Difference between revisions of "Relating taxa to phenotypes"

From phenoscape
(Implications)
(Some vs. All)
Line 18: Line 18:
 
The ''exhibits'' relation has "some of" semantics.  As described above, an ''exhibits'' annotation tells us that some organism or organisms in that taxon possess the phenotype.  This works well for describing characteristics of evolving entities such as taxa.  We can only state what we have seen.  Since membership in a taxon is, ontologically, decided only by genealogy and not by any defining characteristic, it is always possible that a new or undiscovered member of a taxon may exhibit some quality.  Diagnostic characters may epistemologically suggest membership in a taxon, but such characters are not the cause of membership: genus-differentia definitions of taxa cannot be constructed.
 
The ''exhibits'' relation has "some of" semantics.  As described above, an ''exhibits'' annotation tells us that some organism or organisms in that taxon possess the phenotype.  This works well for describing characteristics of evolving entities such as taxa.  We can only state what we have seen.  Since membership in a taxon is, ontologically, decided only by genealogy and not by any defining characteristic, it is always possible that a new or undiscovered member of a taxon may exhibit some quality.  Diagnostic characters may epistemologically suggest membership in a taxon, but such characters are not the cause of membership: genus-differentia definitions of taxa cannot be constructed.
  
These semantics are consistent whether we create annotations on leaf taxa (species) or on higher taxa.  Annotations to higher taxa are exactly the same as those that might be inferred by a reasoner from annotations to its member taxa.  An annotation to a higher taxon tells us only that some member of that taxon possesses the phenotype, but does not provide knowledge that any particular member taxon possesses the phenotype.
+
These semantics are consistent whether we create annotations on leaf taxa (species) or on higher taxa.  Annotations to higher taxa are exactly the same as those that might be inferred by a reasoner from annotations to its member taxa.  An annotation to a higher taxon tells us only that some member of that taxon possesses the phenotype, but does not provide knowledge that any particular member taxon possesses the phenotype.  For example, an annotation entered such as:
 +
* Cyprinidae ''exhibits'' "notched caudal fin"
 +
does not tell us whether its member taxon ''Danio rerio'' has a "notched caudal fin".  It might.
  
 
''To be incorporated: logical inferences from ontological reasoning vs. evolutionary inferences from phylogenetic analysis''
 
''To be incorporated: logical inferences from ontological reasoning vs. evolutionary inferences from phylogenetic analysis''

Revision as of 02:29, 5 February 2009

In Phenoscape, we are annotating the phenotypes of species as described in systematic literature. Because we are working within an ontological framework, we require a well-defined relation to link a taxon to a phenotype. This page describes the properties of, and issues related to, our proposed relation, exhibits.

Defining exhibits

Proposed definition

A taxon X exhibits phenotype Y if phenotype Y has been observed in some organism which is a member of taxon X.

Discussion

In an ideal world, all phenotype annotations would be directly to some organism specimen where the phenotype was observed. However, for various reasons direct specimen annotation is not feasible in practice. So, we create phenotype annotations for "leaf" taxa, species. As the given definition suggests, exhibits statements do not preclude polymorphism within the taxon. Some member organsims of that taxon may have one phenotype while others may have another. For example:

  • Species X exhibits "round dorsal fin"

is not in conflict with another annotation:

  • Species X exhibits "pointed dorsal fin"

Both have been observed within members of this species. Furthermore, phenotypic qualities within PATO cannot really be considered to be alternatives to one another. For example:

  • Species X exhibits "yellow dorsal fin"

would not be supposed an alternative to either of the previous two phenotypes. The situation is no different between "round" and "pointed" themselves.

Implications

Propagation

If a taxon X exhibits a phenotype, this should hold implications for related taxa. Specifically, the parent taxon of X should be inferred to also exhibit that phenotype. This is more easily understood when it is recognized that member taxa do not stand in a subtype or subclass relation to their parent taxa. It is instead purely a genealogical relation. So when one asks, "What phenotypes does the genus Danio exhibit?", the answer is "All the phenotypes exhibited by its member species".

Some vs. All

The exhibits relation has "some of" semantics. As described above, an exhibits annotation tells us that some organism or organisms in that taxon possess the phenotype. This works well for describing characteristics of evolving entities such as taxa. We can only state what we have seen. Since membership in a taxon is, ontologically, decided only by genealogy and not by any defining characteristic, it is always possible that a new or undiscovered member of a taxon may exhibit some quality. Diagnostic characters may epistemologically suggest membership in a taxon, but such characters are not the cause of membership: genus-differentia definitions of taxa cannot be constructed.

These semantics are consistent whether we create annotations on leaf taxa (species) or on higher taxa. Annotations to higher taxa are exactly the same as those that might be inferred by a reasoner from annotations to its member taxa. An annotation to a higher taxon tells us only that some member of that taxon possesses the phenotype, but does not provide knowledge that any particular member taxon possesses the phenotype. For example, an annotation entered such as:

  • Cyprinidae exhibits "notched caudal fin"

does not tell us whether its member taxon Danio rerio has a "notched caudal fin". It might.

To be incorporated: logical inferences from ontological reasoning vs. evolutionary inferences from phylogenetic analysis