PTN002521528 / PTHR24418 NOT issue

[pgaudet]

Hi @huaiyumi @mugity @dustine32

This has been reported in #3304 (comment)

In PTN002521528, FBgn0046685

• has NOT 'protein kinase activity' IKR
• but also getting 'non-membrane spanning protein tyrosine kinase activity' IBA

I think the non-PAINT IKR is being ignored? We had requested that this gets fixed; has this happened ?

@huaiyumi this may be a good discussion topic for the next PAINT call.

Thanks, Pascale

[dustine32]

@pgaudet Right now the qualifier checking is same-term only:
https://github.com/pantherdb/fullgo_paint_update/blob/ba3498c34aa49f79de62c92e83743c032c73b42e/scripts/createGAF.pl#L528
BTW Original ticket for the check: pantherdb/fullgo_paint_update#39

From the perspective of the 'non-membrane spanning protein tyrosine kinase activity' IBA, only all experimental annotations to its same term and same gene are checked. Since there are none of these to compare, there can be no qualifier disagreement, so we think the IBA is good. But we can talk about expanding it to also check experimental annotations to all ancestor terms, which would include 'protein kinase activity'.

The other issue is we don't consider IKR to be "experimental" in this check, so the FlyBase IKR wouldn't be checked even if we did expand to ancestor terms:
https://github.com/pantherdb/fullgo_paint_update/blob/ba3498c34aa49f79de62c92e83743c032c73b42e/scripts/createGAF.pl#L266

Perhaps for the purposes of comparing qualifiers here, we should expand this list of evidence codes to include IKR? Good topic for the PAINT call!

[sjm41]

FYI @hattrill

[mugitty]

The PAINT software supports experimental IKR at leaf node. Furthermore, protein tyrosine kinase activity (GO:0004713) and non-membrane spanning protein tyrosine kinase activity (GO:0004715) is not propagated to node FBgn0046685

[dustine32]

Thanks @mugitty! Another good reason for me to finally finish the XML-to-GAF IBA formatter so that I can use the IBAs directly output from the PAINT tool.

[hattrill]

Hi @dustine32 @mugitty could you explain what is going on? Has a fix been made to the PAINT tool or are you dicussing a pipeline issue here?

[dustine32]

@hattrill Right, I was referring to a pipeline change to make the fix. Specifically, looks like the logic in the PAINT tool (as @mugitty pointed out) is blocking the IBA like we want, and I'm currently working to derive the PAINT IBA GAF files directly from this logic. (short answer: not fixed yet, but will be fixed soon!)

[hattrill]

Thanks @dustine32 I am unconfused now! :-)

[dustine32]

@pgaudet @hattrill While using this ticket's NOT conflict example for developing pantherdb/fullgo_paint_update#52, I noticed this change in AmiGO:

So, the positive IBA to 'non-membrane spanning protein tyrosine kinase activity' seen before has now turned into a NOT IBA due to an IKR added in the PAINT tool.

Just tagging @thomaspd because we were just discussing this strategy and I think he'll approve of it!

[sjm41]

Thanks @dustine32 - that looks good to me from a FlyBase perspective. Does the PAINT 'NOT' only appear in the branch containing fly Wsck? I guess it should do as other members of pthr24418 are bona fide tyrosine kinases.

[pgaudet]

@sjm41 is right - this only applies to a single sequence, I dont think the NOT IBA should be there, the PAINT curator would not have made an annotation here. The presence of a NOT should prevent the positive IBA from being made.

[dustine32]

Thanks for the quick response!

@sjm41 This IKR NOT is on the Arthropoda clade (PTN004704423).

@pgaudet Since the PAINT NOT is an IKR, it will create the NOT IBA. If this IKR (looks like @marcfeuermann created it on 2021-05-05) is changed in PAINT to an IRD it should totally block all IBAs (positive or NOT) for this term to this sequence.

[pgaudet]

OK I see - attaching a new screenshot because this is different from what we used to have #3701 (comment)

[pgaudet]

I think this is now resolved?

[sjm41]

Yes, I think can be closed.