Phenotypes and Diseases are missing taxon, so are many genes

monarch-initiative / monarch-legacy

Monarch web application and API

BSD 3-Clause "New" or "Revised" License

42 stars 37 forks source link

Phenotypes and Diseases are missing taxon, so are many genes #1448

Open jmcmurry opened 7 years ago

jmcmurry commented 7 years ago

Phenotypes and Diseases should have associated taxa. Many genes are also missing taxa. The fact that they do not produces weird search results. I could have sworn that this was already a ticket. It has been mentioned several times in other tickets (eg here), but I'm breaking it out for clarity.

Genes	Diseases	Phenotypes
Taxon tally is too small	No hits in any taxon	No hits in any taxon

jmcmurry commented 7 years ago

For now, I wonder if a stop gap could be that the taxon column disappears if "phenotype" or "disease" facet is selected. @jnguyenx ?

jnguyenx commented 7 years ago

I think that it's expected behavior with the given data. E-commerce websites also have the same pattern if I'm not mistaken. I wouldn't add hacks to hide the underlying problem, but rather fix the actual data.

One improvement I can see is to make the filters with (0) unclickable.

jmcmurry commented 7 years ago

1 to the zero filters being unclickable; however, the problem is not the zeroes themselves, it is the fact that they do not represent the actual data, and fixing the actual data is non-trivial. See also https://github.com/monarch-initiative/monarch-ontology/issues/5

cmungall commented 7 years ago

Taxa for diseases: https://github.com/monarch-initiative/monarch-disease-ontology/issues/59

Taxa for phenotypes: what is expected many not be what is logical here. For example, an ontology term 'abnormality of brain' would be subClassOf in-taxon some Eumetazoa.

If we stratify phenotype ontologies into a lattice as described here: https://github.com/obophenotype/upheno/issues/171

We would have HP be only homo, MP only mouse (or rodents), etc, which people would probably find more intuitive.