UX Feedback on the prototype catalog visualisation

[helenp]

http://snoopy:9000/diagram Accessed 27 November 2019
Comments on existing features, first pass. Nice tool, very fast, these are essentially cosmetic changes and a few new features.

1. Two main routes to diagram visualisation per study and per trait category.
   Suggest have two modes of access arranged as either or - and not both - both is flexible but makes it hard to complete the form. Suggest primary use cases

1.1 By PMID - like this feature, suggest that this could be linked from publication/study page in the catalog - probably also other places too

2. Visualisation circle/stack. I'd suggest we do a short usability test and pick one of these based on that. This may come down to ease of adding the text labels and what happens when we get a huge stack concerns me as the stack may fail to scale more quickly, but rescaling the circles might be simpler.
   2.1 Like the trait visn - suggest have this as a text box matching a free text query as well as the select a high level trait. Fields need re-arranging and I'd suggest that we split the trait piece from the stud(ies) level piece - less confusing, a bit less flexible, but mirrors the search paradigms we offer elsewhere.

3. New Feature suggestions
   3.1 Batch query by PMID, trait or SNP list, or study list - in that order of importance for me

4. Relabelling
   4.1 change the filter diagram to 'generate diagram' make the button much bigger
   4.2 the set of complex filters re: samples, ancestry seem less useful as filters but could work as canned queries - show all assn by ancestry would be a handy default set of queries.
   I don't see the value of sample at this point

[DSuveges]

On the backend a large data frame supports the queries. As this table contains practically the full GWAS Catalog with rsIDs, genes, Accession IDs etc. Expanding the searchable categories, providing batch search is absolutely trivial.

Also it is similarly trivial to return the number of studies instead the number of associations or number of traits. As long as the returned data has the same shape, the visualization is not affected.

I'll address the other raised issues as well.