Closed ljwh2 closed 3 years ago
Testing 2021-06-14:
Oddly, for the example study above with the incorrectly annotated associations (http://snoopy.ebi.ac.uk:9580/gwas/studies/GCST90013538), when I click on the individual variant links, the resulting Variant pages do not contain any studies or associations (e.g. http://snoopy.ebi.ac.uk:9580/gwas/variants/rs6017006)... So maybe there is something else going on with this study
@eks-ebi good catch! data in DB and in Solr Search server is correct, it's UI thing, I'm investigating. Update: Variant page is broken cause it relies on Slim Solr which is outdated on DEV 1. It's back up now (using prod Slim Solr). As for the association's traits, issue was found in DB.
Investigated the issue with the study being split and associations are not. The issue is that this Study was not provided in the Excel sheet that contains the Studies to split. So it must've been split by a curator, but only for Studies, not for associations (background traits for associations is a new addition of this Splitting Issue).
@eks-ebi The study that has split studies but not associations was a submission, I wonder if that had some effect? Do you have any other examples of submissions with a background trait?
Overall functionality looks great!
I agree scientifically it would be better to exclude background traits by default if possible (unchecked box) BUT this requires us to define behaviour of traits which are only background (note - an example of this is Trypanosoma cruzi seropositivity). We can discuss, but on balance it may be better to leave as is.
I noticed an issue for traits that have both child terms and background traits. Unchecking background traits seems to exclude child traits as well. Unchecking and rechecking child traits adds them back in. Example: hypertension EFO:0000537 755 associations, 91 studies. Uncheck background traits: 394 associations, 33 studies. Uncheck child traits as well: 394 associations, 33 studies. Recheck child traits: 515 associations, 58 studies.
Another example is schizophrenia EFO:0000692, behaviour is the same.
Cheers, well spotted! I have deployed a fix.
@ala-ebi Re-tested checking and unchecking child & background boxes, works as expected now.
We discussed and decided it would be better to have the background traits box unticked by default.
'Apply efo traits to associaitons' to associations now takes into account background traits as well, they can be seen later on in the association page
Tested UI again today: I've looked through all the different pages (Trait, Study etc.) and everything looks correct. Ticking/unticking the bg trait box working fine.
Not sure if anything else specific needs testing?
Testing the Curation interface:
The 'Apply efo traits to associations' button works well:
@ala-ebi let me know if there's anything else that needs testing
We discussed adding a tooltip for the background trait field in the UI.
Proposed text: "A trait that is not directly analysed in the GWAS, but is shared by all study participants as a common characteristic." @ljwh2 Does this look ok?
Yes I'm happy with that @eks-ebi
filled background traits in 'All studies v1.0.3' downloads
@ala-ebi I've checked 'All studies v1.0.3' download and all of the background traits seem to be displayed correctly.
I've also checked the tooltip mentioned above and it looks good on the Study page. But I think it would be good to also display the same tooltip in other places where the term "background trait" appears:
In the column header in the Associations and Studies tables:
On the Trait page next to the background traits checkbox
@ljwh2 Would you agree?
done
FAQs now updated to explain background traits, committed to documentation branch: https://github.com/EBISPOT/gwas-ui/blob/documentation/goci-interfaces/goci-ui/src/main/docs/faq-content.adoc
in road map due April