Redundancy in extension default display (modifications)

[ValWood]

I think we could improve the display here quite a lot very easily The red strikeouts seem to be completely redundant with each other The green strikeouts are curation improvements I will look into which will collapse further.

https://www.pombase.org/gene/SPAC24B11.06c

https://docs.google.com/presentation/d/1wEa9FSDwpEgFDpncE-7cxDoTQqKb3EO-n7y6owyOwy0/edit?usp=sharing

[ValWood]

It seems that we don't implement the same rules for modifications that we implement elsewhere?

[ValWood]

Although I thought we had???

[mah11]

It is the same as for other ontologies -- in that ontology paths are only followed for the annotated term, not (yet) for terms used in extensions. The annotations in the screenshot are all to the same term, MOD:00048.

[ValWood]

Ah OK I thought exact match extensions (or presence/absence) were collapsed, but that different specificities of extensions were not understood?

[ValWood]

for example (ase1)

GO:0005515	protein binding	1037
binds peg1 at mitotic spindle midzone	IPI	Bratman SV et al. (2007)
binds klp9 during mitotic anaphase	IPI	Fu C et al. (2009)
binds klp9 part of microtubule sliding involved in mitotic spindle elongation active form anaphase spindle elongation protein 1 unmodified form during mitotic anaphase	IPI	Fu C et al. (2009)

collapses to

protein binding   | binds peg1 at mitotic spindle midzone   | binds klp9 part of microtubule sliding involved in mitotic spindle elongation during mitotic anaphase

in the default view binds klp9 during mitotic anaphase | IPI | Fu C et al. (2009) is hidden?

[mah11]

a lot of extensions are getting collapsed for the sty1 MOD:00048 annotation ... the only one that looks like it should be hidden but isn't is the plain "added by wis1"

[ValWood]

why not also removed by pyp1 redundant with modified residue Y173 removed by pyp1

I was thinking modified residue Y173 added during cellular response to salt stress and modified residue Y173 present during cellular response to salt stress

too but that would need to subsume present_during into added_during (I hope added during is a descendant of present _during)?

I thought the extension filtering could be low priority, but it would be nice as we start doing more modelling of modifications and substrates to present a more concise view.

[mah11]

I hope added during is a descendant of present_during

At present there's no structure to the modification extension relations; it's basically a flat list even though we have it in obo format.

[mah11]

oh yeah, I missed the 'removed by pyp1' thing

[ValWood]

I will get more collapsing by seeing if I can standardize the annotation a bit. GO has a zillion ways to describe the osmolarity response...

[kimrutherford]

I think we could improve the display here quite a lot very easily

Could you add a link to that page?

[kimrutherford]

It seems that we don't implement the same rules for modifications that we implement elsewhere?

It should be the same for all ontologies.

[ValWood]

sorry I meant to do that: https://www.pombase.org/gene/SPAC24B11.06c

[kimrutherford]

sorry I meant to do that: https://www.pombase.org/gene/SPAC24B11.06c

Thanks. I'm surprised the redundancy removal code has this bug as it's been in use for a year or two now and it seemed to work well. It's too complicated to look at on a Friday night. I'll have a look on Monday.

[ValWood]

There is no hurry ;) Nothing is presented incorrectly. I wouldn't really call it a bug.

[kimrutherford]

I wouldn't really call it a bug.

I'm just a bit concerned that this is a symptom and there are deeper problems. I'll have a look next week.

[kimrutherford]

I'm just a bit concerned that this is a symptom and there are deeper problems.

I've dug into this now. It was happening because the "modified residue" isn't really part of the extension. It gets inserted as a fake extension part for display purposes but was added too late in the process. That was causing the summarisation code to mess up. It's fixed I think and we'll see the results on Monday morning. I'll check the main site then.

[kimrutherford]

I think it's fixed. It's quite complex code though so I can't guarantee that there aren't other issues.

Let me know if you see anything else.

[ValWood]

on

O4'-phospho-L-tyrosine

modified residue Y173 removed by pyp1 should be hidden by modified residue Y173 added by wis1 added during cellular response to salt stress removed by pyp1

[kimrutherford]

should be hidden by

This has turned into a can of worms. :-)

I can't fix this problem without causing different bugs. I'm going to leave this for a while and possibly do a rewrite to fix this and to remove redundancy using the hierarchy of terms in extensions: https://github.com/pombase/website/issues/1620#issuecomment-708154147 https://github.com/pombase/website/issues/1619#issuecomment-708005613

[ValWood]

No worries, no hurry becasue it isn't incorrect as it is.

. Better to do it all together, no point in tinkering.

[ValWood]

related https://github.com/pombase/pombase-chado/issues/747

[ValWood]

here is an example on process:

Mal3

I'll add a few different types for checking.

[ValWood]

I noticed these today on the Cdc2 page, I thought these would be a single set?

[kimrutherford]

I noticed these today on the Cdc2 page, I thought these would be a single set?

It looks like in this summary row the "phosphorylates" is a "has_input" relation:

phosphorylates bit61, cdc15, cdc23, crf1, csi1, dbl7, dfp1, dnt1, drc1, fkh2, ies4, iss10, lem2, mdb1, mis4, msh6, mvp1, orc1, rad2 during mitotic G2 phase

And in this row, the "phosphorylates" is a "has_direct_input":

phosphorylates rum1 during mitotic G2 phase

I thought I had fixed this problem: https://github.com/pombase/website/issues/1701#issuecomment-842108424 but must have got it wrong. I'll fix it.

[kimrutherford]

I noticed these today on the Cdc2 page, I thought these would be a single set?

I've fixed that so now they are all in one row of the summary:

phosphorylates bit61 , cdc15 , cdc23 , crf1 , csi1 , dbl7 , dfp1 , dnt1 , drc1 , fkh2 , ies4 , iss10 , lem2 , mdb1 , mis4 , msh6 , mvp1 , orc1 , rad2 , rum1 , sec16 , sgo2 , sld3 , snt1 , ulp2 during mitotic G2 phase

[ValWood]

much better!

[ValWood]

pdb1

I think this one is because the modified forms are not handled by the gene name:

[kimrutherford]

Should it be this instead?:

   binds hhf1/Me3:(K20), hhf2/Me2:(K20), hhf3/Me3:(K20)

[ValWood]

comment on pdb1 methylated histone binding is a parent of H4K20me3 modified histone binding. I think we need to explicitly add the target gene name in addition to the modified form. Although we would like this to be 'inferred' . So actioning this ticket https://github.com/pombase/website/issues/1710 would resolve this particular extension problem.

[kimrutherford]

Hi Val.

How much redundancy is there in the current display? It's changes a bit since you created this issue?

https://www.pombase.org/gene/SPAC24B11.06c

[ValWood]

That one looks fine. I closed off everything that is no longer an issue. The "methylated histone binding" eg is still current

[ValWood]

https://www.pombase.org/gene/SPAC24B11.06c looks fine. I think this ticket can close?

[kimrutherford]

I think this ticket can close?

If you're happy with it, I'm happy. :-)

[ValWood]

Yep, if there is anything else it is very minor and we can open a new ticket