Investigate: getting no records from Automat APIs?

[colleenXu]

When analyzing automated test runs, I've noticed a lack of edges/records from Automat APIs - and that seems kinda odd.

We'd need more investigation to see if Automat is truly returning no records for subqueries (and if that's an Automat problem) or if there's an issue processing them into records...

Ex: overall PK 7d6e623f-d7c5-4321-9dd2-64da17107a01, BTE PK 92179571-10b6-4128-ae79-081592a9be85. This is the "decreases CYP2D6" query for the Automated test run 7/14. Automat APIs were only queried in template 4 (does that make sense?) and no records found (does that make sense?)

---

Update: I have seen cases where Automat APIs are contributing records to the results, while reviewing the 7/7-7/14 runs. So it's not like we're NEVER getting records from these APIs.

[colleenXu]

Side note: in 7/7 automated test runs, there may have been a problem with the Automat registrations/yamls. BTE raised errors and didn't run the MVP2 templates 4-5 for the "decreases CYP2D6" as a result. Overall PK 2216ddd9-3a09-4a34-afc5-38416798d4ac, BTE PK c1c9270a-2bec-491d-8daf-ad8ce2c30400

~Update: this happened in the 7/14 automated test run too "semaglutide increases" (main PK: 76cbced8-e9f4-4271-9599-b6d3d2d81d6d, BTE PK: b8d4fd40-aad9-4d3d-a382-017892e98076)~

[colleenXu]

Here's my thoughts so far:

• I think Automat edges aren't showing up in MVP1/"treats" responses as much as I would expect.
• Automat edges are maybe not showing up in some MVP2/"chem affects gene" responses. But it's obfuscated with the TypeError/sub-queries not being made. That's the linked issue above.

[colleenXu]

Probably also affected by the dropped edges fixed in matching vs non-matching "direct" edge handling.

Will need retesting to see if this is still a problem...