want feedback on NTR to PATO

[dsiegele]

During the Phenotype RCN meeting in February 2016, I talked with Chris about the idea of having PATO use 'altered' instead of 'abnormal.' Chris told me to post the proposal on the PATO issue tracker, which I did, but never got any feedback from anyone (https://github.com/pato-ontology/pato/issues/87).

I'd like to submit NTRs to PATO, but want people's feedback both about doing that and also about the terms themselves. Here are the terms as they currently are in OMP.

id: OMP:0007189 name: altered_relative_to def: "A quality which inheres in a bearer that is altered relative to the quality that inheres in a designated control entity." [OMP:DAS] is_a: PATO:0000068 ! qualitative

id: OMP:0007188 name: altered_process_quality def: "A quality which inheres in an process that is altered in some way relative to a designated control." [OMP:DAS, PATO:0001236] is_a: PATO:0001236 ! process quality relationship: has_quality: OMP:0007189 ! altered_relative_to intersection_of: PATO:0001236 ! process quality intersection_of: has_quality OMP:0007189 ! altered_relative_to

[pbuttigieg]

Relation (to some norm) type qualities are tricky. They read reasonably enough, but I find it uncomfortable to say that they inhere in one entity. In a way, I'd rather work with RO object properties like "increased pathogenicity relative to" and link the entities in question, to reduce ambiguity. That RO class would be linked to a more clear quality (or disposition, to be more accurate in BFO terms) like "pathogenic " or similar.

[dsiegele]

Pier,

While building OMP we realized that there are two classes of phenotypes: independent phenotypes and relative phenotypes. An independent phenotype is an observation of a strain in a particular environment that is made without reference to anything else. A relative phenotype is an observation of a strain in a particular environment where the phenotype is described relative to a specified reference phenotype. OMP has different terms for the two classes of phenotypes, and the OMP annotation system allows you to link the two annotations. In our web-based annotation system, each annotation is automatically given a unique identifier (OMP_AN:xxx).

It's easiest for me to explain this using the example of motility phenotypes. If a paper reports that E. coli in a particular environment is motile that would be an independent phenotype, and it would be annotated with OMP:0000005 'presence of cell motility.' If the same strain of E. coli in a different environment is nonmotile, it would be annotated with OMP:0000133 'absence of cell motility.' These are both independent phenotypes because the observations are made without reference to anything else.

However, in some cases we might want to compare two phenotypes. To do this, one phenotype is designated as the reference, and the other phenotype is described relative to the first one. In my example, E. coli in motility buffer at 30C would be motile and would still be annotated with OMP:0000005 'presence of cell motility.' E. coli in motility buffer at 42C would be less motile or perhaps even nonmotile. Because we want to compare this phenotype to the initial observation we would annotate it using the relative phenotype term OMP:0000002 'decreased cell motility.' The annotation of the second phenotype would include the annotation accession for the designated reference.

You can see examples of this in annotations on the OMPWiki. AW518 is a mutant strain that lacks the ability to respond to certain repellents. These phenotypes are relative to the behavior of the "wild-type" parent strain AW405.

If you look at http://microbialphenotypes.org/wiki/index.php/OMP_ST:773_!_Escherichia_coli_K-12_AW518, you can see that annotation OMP_AN:341 shows that strain AW518 has decreased negative chemotaxis to acetate. The 'relative to' field gives the reference annotation (OMP_AN:403) and a link to the web page with the annotation. The 'relative to' field also shows a comparison of the two genotypes and the two environments. In this case we see that there is a genetic difference between the two strains but the experimental conditions are the same.

This is one of the reasons why I think PATO needs a quality 'altered_relative_to.' I hope I was clear. Another reason is related to the way that many PATO terms make reference to a normal or average quality. Here are 2 examples:

abnormal PATO:0000460 A quality inhering in a bearer by virtue of the bearer's deviation from normal or average.

decreased process quality PATO:0002302 A quality of a process that has a value that is decreased compared to normal or average.

Because OMP is meant to be used for any microbe there isn't a single normal or average reference to be used as the basis of comparison. The reference will always needs to be designated. Moreover, in some cases we might want to capture the phenotypes of two strains where neither is 'normal or average,' such as the case of a mutant strain and a pseudorevertant.

I look forward to further feedback.

Regards,

Debby