Lookup Edges in Support Graphs

[sstemann]

In the Test UI > MVP1 > What drugs may treat common cold: https://ui.test.transltr.io/main/results?l=Common%20Cold&i=MONDO:0005709&t=0&r=0&q=a49c78af-b108-490e-b48f-f7aa7a5e568c

It looks like there is a Lookup in the Inferred Paths section

I checked the ARAs in the ARAX GUI and it looks like these edges, are coming from an ARAGORN support graph (based on the EPC - repodb and drug central) - should lookups be represented in support graphs? these edges also appear in the result graph

Result graph:

Support graph:

[gprice1129]

Added it to our backlog for prioritization

[gprice1129]

@sstemann there is nothing stopping an ARA from adding a lookup edge as part of the supporting paths for an inferred edge. I'm also not sure why we would want to explicitly disallow this behavior. Can you give more context on why you think this is wrong?

[sstemann]

we discussed at the 2/21 UI call. UI team to reach out to @cbizon and @uhbrar

[cbizon]

So what's happening here is that we have some rules that are a combination of paths. And in particular, we have quite a few that say something like: If A-ameliorates-B and (Other specific 2-hop path connecting A and B) then A-treats-B. So the fact that we have a paper saying something vague (ameliorates) about A and B, plus a more mechanistic path from A to B adds up to say treats.

Now, the tricky part is that when we run that rule, treats is a sub-predicate of ameliorates. So A-treats-B comes back from KPs when we ask for A-ameliorates-B.

It's unintended, and I can see why it's confusing. We're going to have to work on these rules soon anyway when we deal with the treats refactor next month, so my thought would be to wait until then to handle this unless it's felt to be a bigger problem.

[sstemann]

i'm adding another example, what chemicals may down regulate VKORC1, with the 5 lookup paths in Lookup and multiple lookup paths in Inferred - its really hard to understand why there are so many different paths:

https://ui.test.transltr.io/main/results?l=VKORC1%20(Human)&i=NCBIGene:79001&t=2&r=0&q=91b14094-a555-492f-973c-df9103c2c091

[sstemann]

@cbizon when is it scheduled to be handled?

[cbizon]

We have it scheduled for post-treats. But that's basically now. I think we should target Eel unless you think it's more urgent, in which case we can shoot for Octopus.

[sstemann]

the common cold example is OBE in Prod/Fugu. https://ui.transltr.io/results?l=Common%20Cold&i=MONDO:0005709&t=0&r=0&q=ec00ed3a-b1a3-4c78-bfb7-9a9ef2a3a309

the decrease VKORC1 example, still exists in CI https://ui.ci.transltr.io/results?l=VKORC1%20(Human)&i=NCBIGene:79001&t=2&r=0&q=1eba9535-d9e8-4fcd-ae50-6edb89acdef8

the decrease VKORC1 example, sort of exists in Prod - all the edges are Direct

so not sure what is going to happen with the Guppy release

[sstemann]

post Guppy deployment to Test, this issue still occurs.

https://ui.test.transltr.io/results?l=VKORC1%20(Human)&i=NCBIGene:79001&t=2&r=0&q=29cda046-a299-4e79-b6e8-3c16e4827598

The inferred edge lists on ARA - ARAGORN