Inconsistent display of 'subclass' edges in support paths

[mbrush]

Search on test for what may treat breast cancer: https://ui.test.transltr.io/main/results?l=Breast%20Cancer&i=MONDO:0007254&t=0&r=0&q=a6d28087-f193-441c-ade3-26b04fdd0dd1

Get Doxyrubicin as a result that says "4 paths that treat her2 positive breast carcinoma".

Issue 1: 'her2 positive breast carcinoma' is not the disease I searched for . . . it is clearly a subtype of 'Breast Cancer, so it makes sense that the Doxyrubicin result shows up. But in many cases the hierarchical relationship between diseases may not be clear, and the user could do with some explanation of why diseases they didn't search for are showing up in their result.

This is also an issue in the Paths for this result, as the disease at the end of each support path is different for every path.

Again, these are all subtypes of the 'Breast Cancer' query term - which may be fine (but the user could do with some explanation of why these paths came back).

Issue 2: The main issue that concerns me however, is that if I look at another result (Mitoxantrone), I see subclassOf edges in support paths, when a subtype of the queried term (Breast Cancer) shows up there:

This begs the question: why are these subclass of edges showing up in support paths here, but not in the Doxyrubicin result support paths?

I am guessing this is a difference between how Unsercret Agent (the source of the Doxyrubicin prediction) vs how BTE (the source of the Mitoxantroneprediction) are handling things when subclasses of the queried disease show up as a Result or Paths - which leads to inconsistencies noted above in the UI.

[sstemann]

the ARAs are not sending subclass edges in the result graphs for Doxorubicin

ARAX - Treats edges, no subgraphs

Unsecrets Edge Support Graph is also for the resultant disease without a subclass of edge

[sierra-moxon]

from TAQA:

• we decided that we wouldn't have folks implement the subclasses in the aux graph yet, but one team is ahead of the gang.
• to fix, we need all ARAs and KPs to start doing the way we came up with in architecture - large fix.
• can't do it this month, but put this fix in the "parking lot" for next month/next cycle.

[sierra-moxon]

@capasfield - this is the ticket for subclassing discussion in architecture :) -- if I assign you then it falls off the "needs review" list :) -- just assigning so that it makes it to the architecture discussion.