uPheno currently has two top-level classes that both (roughly) mean "a change in a trait or biological attribute":
- UPHENO_0001001 (phenotype)
- UPHENO_0001002 (phenotypic effect)
It's unclear why both exist. Running a SPARQL query for descendants of UPHENO_0001001 that are not also descendants of UPHENO_0001002 returns a small set of ~30 classes. These fall into roughly three categories:
- "Normal" phenotype classes — e.g. "normal phenotype", "no abnormal phenotype detected; wild type", "normal molecular function"
- General grouping classes — e.g. "craniofacial/craniocervical phenotype", "mortality/aging phenotype", "vision/eye phenotype"
- Miscellaneous terms — e.g. "haploinsufficient", "haploproficient", "auxotrophy"
The apparent reasoning for excluding these from "phenotypic effect" seems to be that normal states and broad groupings aren't "effects." But this distinction doesn't hold up: observing a normal phenotype is still an observation of a phenotypic state, and grouping classes are just organisational containers for effects. There's no meaningful semantic boundary here.
Proposal
- Deprecate UPHENO_0001002 (phenotypic effect) — it adds no value as a separate class. Since it's a subclass of UPHENO_0001001, all its descendants are already under UPHENO_0001001 and nothing needs to be reclassified.
- Rename UPHENO_0001001 from "phenotype" to "phenotypic effect".
- Adopt the following definition for the renamed UPHENO_0001001:
phenotypic effect — An observable or measurable characteristic of an organism, whether normal or abnormal, resulting from the interaction of its genotype with the environment. This includes deviations from typical trait values, the absence of such deviations, and organisational groupings of related phenotypic observations.
Why "phenotypic effect" over "phenotype"?
"Phenotype" is broad and sometimes used loosely to refer to the entire phenotypic profile of an organism. "Phenotypic effect" more clearly communicates that each instance is a specific, discrete observation about a trait — which is what uPheno classes actually represent.
@sbello Can you coordinate this issue with the editors? I am going to proceed under the assumption that one way or another we CAN merge the two terms, so I can proceed with my work on aliging uPheno with COB.
uPheno currently has two top-level classes that both (roughly) mean "a change in a trait or biological attribute":
It's unclear why both exist. Running a SPARQL query for descendants of UPHENO_0001001 that are not also descendants of UPHENO_0001002 returns a small set of ~30 classes. These fall into roughly three categories:
The apparent reasoning for excluding these from "phenotypic effect" seems to be that normal states and broad groupings aren't "effects." But this distinction doesn't hold up: observing a normal phenotype is still an observation of a phenotypic state, and grouping classes are just organisational containers for effects. There's no meaningful semantic boundary here.
Proposal
Why "phenotypic effect" over "phenotype"?
"Phenotype" is broad and sometimes used loosely to refer to the entire phenotypic profile of an organism. "Phenotypic effect" more clearly communicates that each instance is a specific, discrete observation about a trait — which is what uPheno classes actually represent.
@sbello Can you coordinate this issue with the editors? I am going to proceed under the assumption that one way or another we CAN merge the two terms, so I can proceed with my work on aliging uPheno with COB.