Messenger RNA Evidence for MODY

[sstemann]

https://ui.test.transltr.io/main/results?l=Maturity-onset%20Diabetes%20Of%20The%20Young&i=MONDO:0018911&t=0&r=0&q=f8de3997-f479-49c7-b9ed-b15dc9db9ae4

the messenger RNA answer, via the UI, only has evidence "Unsecret Agent"

in the ARAX GUI the Support Graph says Genes are part of the inference, maybe Genes was general and blocked but the "lookup path" remains? In general though, isnt this just too generic?

[cbizon]

Seems like 2 issues:

1. Add Messenger RNA to block list (@MarkDWilliams ?)
2. The provenance here shouldn't say Unsecret I think. Not sure if this is the underlying data or the UI (@webyrd , @kaiwenho can you comment please?)

[cbizon]

I also think we should add "Promoter" and "Double Stranded Rna" to the block list (found in MODY results)

[kaiwenho]

I agree with @sstemann that 'Genes' was general and blocked. Messenger RNA -treats-> Maturity-onset Diabetes Of The Young is the inferred edge generated by Unsecret, with the supporting edges Messenger RNA -affects-> Genes Genes -causes-> Maturity-onset Diabetes Of The Young and both supporting edges are coming from semmeddb. I am assuming these two 'bad' edges were blocked by ARS, since in the ARAX GUI the supporting edges are still there, but they are not in the UI. However, the inferred edge has not been blocked/removed even all (4 in total) supporting edges were gone (/blocked).

[cbizon]

@MarkDWilliams if all of the paths are removed, shouldn't the overall result be removed as well?

[cbizon]

Also add "Response Elements" which I found in running "treats Marfan Disease"

[cbizon]

Also "Elements"

[sierra-moxon]

copper atom too?

[sierra-moxon]

rerun results: https://ui.test.transltr.io/main/results?l=Maturity-onset%20Diabetes%20Of%20The%20Young&i=MONDO:0018911&t=0&r=0&q=81aba51f-2b8f-46b7-9b80-da71caadac4a

[sierra-moxon]

from the relay: Many of the "rerun results" above should be on the block list. If a result shows as an inferred path but contains a blocked item, then ARS will return the blocked edge; if ARS returns all evidence, then it removes all of it but isn't touching the primary knowledge source edge, which then shows up as a lookup (cause no support graphs)...

todo: just get rid of everything - no bare assertions without support graphs.

[MarkDWilliams]

I believe I have a fix for this now. Just to make sure we're all on the same page about it, with this fix

• If all of the support graphs on an edge get removed for blocklist reasons, we remove that edge
• If that edge is part of any analyses, it is removed from there as well
• If removing edges from an analysis leaves us with an analysis that has no edges, we remove that analysis
• If removing that analysis leaves us with a result which has no analyses, we remove that result.

All said, just to give an idea of numbers it looks like these are the differences in the number of things removed from Unsecret's merged response pre vs post fix: Nodes: 25 -> 25 Edges: 289 -> 386 Results: 24 -> 105 Aux graphs: 128 -> 128 Analyses: 16->99

I have not yet added all the excellent blocklist additions mentioned in this thread yet (I will do that next), but just as a heads up that the numbers above are likely to get larger

[cbizon]

This sounds good to me!

[sierra-moxon]

retested on test this morning; seeing copper atom, etc. in the results. For messenger RNA, I still see the one lookup path.

[sstemann]

trickle up is working in Test!