Getter for json output for discrete (categorical) value columns

[jarmoza]

Closes #406
See also #408

• New getter introduced in store getCategoricalJsonOutput that takes a column name and outputs a json object with annotated values for that column that is compliant with the schema as noted in getter to create JSON representation of discrete column #406
• New unit test file store-getter-getCategoricalJsonOutput.cy.js with tests for schema compliance and value/value formatting checks for the 'IsAbout' and 'Levels' sections of the json output

[surchs]

ah, didn't realize you had already picked this up @rmanaem.

[rmanaem]

@surchs No worries

[rmanaem]

Thanks for the PR @jarmoza! I requested a minor change in one of the test cases, let me know what you think.

[surchs]

Just mentioning a couple of things I noticed here:

• this was missing from the spec, but we have a process for "missing values" and they should be part of the output data dictionary. See https://www.neurobagel.org/documentation/dictionaries/#assessment-tool for example.
• what is stored under valueMap in the store is not the text-label of a categoricalOption, but the unique identifier. See the relevant component test for an example -> we have the identifier, and we need to lookup the human label (not the other way around)
• although it is true that currently all variables a column can be "isAbout" live in the neurobagel namespace, not all of them will have a unique identifier that is an exact match with their human readable label. I see @jarmoza has already made an issue to align the label with the identifier: Swap 'Subject ID' for 'Participant ID' #425 However I think this will quickly get hard to keep up. I think a better way is to do what we've done for the (currently hardcoded) categorical options and add an identifier attribute to the pre-defined categories (e.g. here). That will allow us to
  • have the identifier clearly spelled out in a single place - reducing mental math to parse it from the code
  • have a variable mapped to a different identifier than it's human label (e.g. something with a space "Participant ID" could map to http://neurobagel.org/vocab/ParticipantID)
  • change to a different, non neurobagel namespace, ...

[surchs]

Thanks a lot @rmanaem for taking on the PR! I read through the whole thing and left some comments. I think the missingValues implementation is assuming the wrong data structure for missingvalues (it's an array, not an object). Could you fix that and take a look at some of the other comments (:cherries: are just nice-to-have's you can ignore).

[surchs]

Nice, looks good @rmanaem. Good to merge 🧑‍🍳

Thanks @jarmoza and @rmanaem for addressing this!