Allow (with warning) experimental and ancestral (IBD) annotations that violate taxon constraints

[dustine32]

Hi @mugitty!

Creating this ticket for tracking how we want the PAINT curation tool and validator to respond when taxon constraints are violated (TC invalid). The requirements:

1. Display TC invalid experimental GO annotations in the PAINT tool
2. Allow TC invalid experimental GO annotations to be supporting evidence for IBD annotations
3. Allow TC invalid IBD annotations to be created by curator and retained (considered valid) by the validator tool
4. Create TCVs to block propagation of IBAs

Warnings should be output/logged.

Warning for requirement 2:

PTHR10997 Warning Experimental evidence IGI,IMP to PTN000108260 for annotation 385617457 to term GO:0016203 with species Drosophila melanogaster violates taxonomy constraints.

Warning for requirement 3:

PTHR11878 IBD to PTN000933900 for annotation to term GO:0098794 for species Metazoa-Choanoflagellida violates taxonomy constraints.

Requirements 2 and 3 above already have these warnings being logged. (I stole these examples from your log) Requirement 4 (creating TCVs) is currently implemented and seems to be working well.

Let me know if you have any questions.

[mugitty]

Requirements 1, 2, and 3 can be implemented.

For requirement 4, until the taxonomy constraints are resolved, the tool should not be creating TCV's. There are two scenarios where TCV's are applicable. One is if the IBD annotation itself should not be created. Two is when the species of a descendant node violates the taxonomy constraint.

As a visual clue to the curators, the experimental annotations in the matrix can be a different color or different shade of green, if there are taxonomy violations. This may compel them to create tickets, if the constraint is incorrect.

[dustine32]

Thanks @mugitty!

For req 4, you're right in that creating TCVs to block IBA propagation is inconsistent with the "taxon constraints should do no harm" approach of reqs 1-3. The fun thing is that, since the GO release pipeline uses the same gaferencer TC logic to filter out IBAs, these IBAs will effectively be blocked downstream anyway. So I guess it's a question of whether or not we at least want these TC invalid IBAs in the PAINT IBA release files, even if they won't make it into a GO release. I have no opinion on this last part. So, we can update the above requirements if we wanna drop req 4.

I like your idea of the visual clue for exp annotations that violate taxon constraints.

[mugitty]

Once the taxon constraints are formalized and taxonomy checks are enabled, PAINT will automatically insert TCV's in descendant nodes and also not permit IBD if the species violates the taxonomy check for the term.

@pgaudet, @huaiyumi is this how you want to proceed with the update?

In the meantime, do you want the taxonomy check to disabled in the PAINT software as well.

[mugitty]

Create menu item that will bring up taxonomy violations.

If user creates an annotation that results in taxonomy violation, bring up text box with link to https://github.com/geneontology/go-ontology/issues and request user to create bug report and tag as taxon constraint.

Also determine if any descendant nodes will violate taxonomy constraint checks. If yes, warn user when ancestor annotation is created