cBioPortal / cbioportal

cBioPortal for Cancer Genomics
https://cbioportal.org
GNU Affero General Public License v3.0
636 stars 487 forks source link

OQL doesn't complain if appropriate data type isn't selected under genomic profiles #3903

Closed tmazor closed 5 years ago

tmazor commented 6 years ago

OQL with EXP complains if expression isn't selected in molecular profiles. This is a good thing. image

However, no other data type acts like this. All of the scenarios below should similarly provide a warning and prevent the user from submitting the query: image

image

image

Also, that error for expression should say "Genomic Profile", not "Molecular Profile"

alisman commented 5 years ago

@jjgao @tmazor would it at all make sense to just shut off the profile selector as soon as we detect someone has typed qualifiers directly into the oql? Just looking for a possible simplification

adamabeshouse commented 5 years ago

@alisman another idea: we could auto-select a profile in that case?

alisman commented 5 years ago

I mean, if these are alternative methods to do the same thing, maybe it's simpler to be exclusive about them. It could avoid confusion to mix and match. I would not expect that things I type into that box automatically toggle a selection up above. On the other hand, maybe it's too much to expect them to type all qualifiers into the field? Definately need product input on this.

adamabeshouse commented 5 years ago

the only issue with hiding it is that, AFAIK, there could be more than one option for mrna profile

tmazor commented 5 years ago

@alisman I don't think we can turn off the profile selector when a user starts typing OQL because a user could want to query genomic profiles that aren't explicitly called out -- for example, I could type EGFR: AMP MUT CDKN2A That would only select mutation & copy number profiles. But I might want to also select mRNA and use the default OQL applied to CDKN2A. So I think we need to maintain user control of the profile selection.

We could consider auto-selecting a profile if OQL is added, but then we run into questions like @adamabeshouse pointed out, for example, which mRNA profile to select if there are multiple. And I agree, it's not the most intuitive to have some in the gene box change a setting above.

That said, I do agree that it would be great to simplify this process. I just don't happen to have a good idea for how to do it...