ongoing flip-flopping of annotation #678

ValWood · 2018-05-09T09:42:38Z

OK I don't know what the issue is but we seem to be loading some version of the ontology which
is different from the main GO obo. Or something is occurring so that some relationships are not recognised. I suspect the former, because this changes on a daily basis.

summary:

The number of annotations to
"conjugation with cellular fusion" keeps changing, even though we are not changing the annotation.

previously 108:
today :101
see the history here
geneontology/go-ontology#15641 (comment)

for some reason, some gene products which are annotated to terms which are descendents of "
"conjugation with cellular fusion" do not slim to "conjugation with cellular fusion".

Here is an example
lsc1 https://www.pombase.org/gene/SPBC530.13

has

GO:1904788 "positive regulation of induction of conjugation with cellular fusion by regulation of transcription from RNA polymerase II promoter"

but this does not always slim to "conjugation with cellular fusion"
(see the GO slim overview section)

ValWood · 2018-05-09T09:44:37Z

and, even though you can "ascend" to this term page:
https://www.pombase.org/term/GO:0000747
from https://www.pombase.org/term/GO:1904788
lsc1 is not listed here

ValWood · 2018-05-09T09:45:02Z

Its odd that sometimes the parent is there and sometimes not though.....

mah11 · 2018-05-09T10:20:22Z

we seem to be loading some version of the ontology which is different from the main GO obo

As far as I can tell, this is not the explanation. Kim says we load go-basic.obo, and when I checked yesterday the version in https://curation.pombase.org/dumps/latest_build/ontologies/ was identical to the go-basic.obo in the GO repository for the same date.

I can check again when Canto returns, but if we were loading different content, wouldn't that affect the "Parents" on the ontology term pages too?

kimrutherford · 2018-05-09T13:14:13Z

I've had a look at this and I'm very confused. We load the transitive closure into Chado and then that is used by the website code to propagate the annotations upward. In Chado for today's load GO:1904788 has GO:0000747 as a parent via regulates. In yesterday's load Chado doesn't have that inferred relation.

(pombase-dev has today's update and the main site is still on yesterday's)

We create the transitive closure with owltools. I ran it on the command line for today and yesterday's go-basic.obo files and got different results.

For today it reports:

GO:1904788   regulates       5       GO:0000747

(The 5 is the length of the path that owltools found but 5 not the shortest path - odd)

Yesterday's output doesn't have that line.

Strangely, both today and yesterday have:

GO:1900406   regulates       3       GO:0000747

and GO:1904788 is_a GO:1900406

Perhaps there is a bug in owltools?

I'll dig deeper tomorrow after a sleep. I'll download the latest owltools to see if that makes a difference.

mah11 · 2018-05-09T13:21:45Z

That is bizarre. The go-basic.obo files from the last 4 builds have identical paths involving GO:1904788 (3 of the 4 are completely identical, and the only difference from the most recent file is that two MF terms got merged).

ValWood · 2018-05-09T14:05:28Z

I knew there was a problem, but it isn't super urgent to fix, but we should get to the bottom of it......

kimrutherford · 2018-05-10T02:52:20Z

That is bizarre. The go-basic.obo files from the last 4 builds have identical paths involving
GO:1904788

Yep, that's strange. I've just tried the latest version of OWLTools and the results are the same. Is there any else we can try or check before I create an owltools issue?

mah11 · 2018-05-10T09:01:57Z

I'm pretty much stumped. It sounds like you can feed the same file to owltools and get different results on different days. I got nuthin'

ValWood · 2018-05-10T10:35:24Z

OK just more spooky craziness!

ValWood · 2018-05-14T15:07:03Z

@cmungall says can you open a ticket on owl tools tracker and tag him he will take a look

mah11 · 2018-05-14T15:09:32Z

guessing that Chris means OWLTools - https://github.com/owlcollab/owltools/issues

ValWood · 2018-05-14T15:14:23Z

yep.
I would do it but I'm not sure how to describe the problem

cmungall · 2018-05-14T15:57:50Z

it's ok let's try and sort out the problem here, but in general you have a higher chance of getting my attention on another tracker. Unfortunately the current owltools developers' last day is today but I can handle this.

@kimrutherford can I confirm that your pipeline does something like:

owltools go-basic.obo --save-closure-for-chado foo.tsv

kimrutherford · 2018-05-14T22:24:39Z

owltools go-basic.obo --save-closure-for-chado foo.tsv

Yep, that's what we run.

These are the OBO files I was testing in my comment above - #678 (comment). The output of owltools changed quite a lot between these two files but we can't see why:

https://curation.pombase.org/kmr44/go-basic-2018-05-07.obo
https://curation.pombase.org/kmr44/go-basic-2018-05-08.obo

ValWood · 2018-05-15T14:26:16Z

look, this was yesterday

and actin cytoskeleton org was 128 yesterday

we did not change anything

it flip-flops like this every couple of days..... I really think it is a difference in the GO file

mah11 · 2018-05-15T15:52:41Z

But there aren't any differences in the GO files that would account for changing counts of GO:0000747 annotations. Literally the only difference between the go-basic.obo files from yesterday's and today's builds is a typo correction in a synonym of a MF term. Two letters transposed. Sometimes the GO files have been identical.

For the files Kim linked above (go-basic-2018-05-07.obo and go-basic-2018-05-08.obo) this is the complete diff:

2c2
< data-version: releases/2018-05-04
---
> data-version: releases/2018-05-07
265459c265459
< name: 2-octaprenyl-6-methoxy-1,4-benzoquinone methyltransferase activity
---
> name: 2-octaprenyl-6-methoxy-1,4-benzoquinone methylase activity
265460a265461
> alt_id: GO:0102005
265463c265464,265466
< synonym: "2-octaprenyl-6-methoxy-1,4-benzoquinone methylase activity" EXACT []
---
> synonym: "2-octaprenyl-6-methoxy-1,4-benzoquinone methyltransferase activity" EXACT []
> xref: EC:2.1.1.201
> xref: MetaCyc:2-OCTAPRENYL-METHOXY-BENZOQ-METH-RXN
438631,438639d438633
< id: GO:0102005
< name: 2-octaprenyl-6-methoxy-1,4-benzoquinone methylase activity
< namespace: molecular_function
< def: "Catalysis of the reaction: 6-methoxy-2-octaprenylhydroquinone + S-adenosyl-L-methionine <=> H+ + S-adenosyl-L-homocysteine + 5-methoxy-2-methyl-3-octaprenylhydroquinone." [EC:2.1.1.201]
< xref: EC:2.1.1.201
< xref: MetaCyc:2-OCTAPRENYL-METHOXY-BENZOQ-METH-RXN
< is_a: GO:0008168 ! methyltransferase activity
<
< [Term]

But Kim gets different results form OWLTools for the 05-07 vs 05-08 files.

ValWood · 2018-05-16T15:19:03Z

So strange, any ideas @cmungall ?

ValWood · 2018-05-16T15:21:07Z

and back again today??????

ValWood · 2018-05-31T11:02:18Z

Summary, transitive closure was inconsistent over the same input file.
No alternative....

kimrutherford · 2018-06-05T02:43:23Z

I've made an owltools issue about this: owlcollab/owltools#256

ValWood · 2018-06-05T03:00:31Z

good, it still happens every couple of days...

2 june

3 June

ValWood · 2019-04-18T13:11:43Z

Ah this is the same issue that I just put on the GO tracker. I thought this was all done and dusted....
https://github.com/geneontology/go-ontology

kimrutherford · 2020-10-14T01:40:06Z

Hi Val.

The load worked last night including the improvements from geneontology/go-ontology#17171 (comment)

For now I haven't updated the main site so it still shows the load from Monday night. http://dev.pombase.kmr.nz has the updated results from Tuesday night's load which I think are an improvement. I'll hold back the update until you've had a look.

Looking at the BP slim, the counts for actin cytoskeleton organization are 99 vs 101:
https://www.pombase.org/term_genes/GO:0030036
http://dev.pombase.kmr.nz/term_genes/GO:0030036

mitotic cytokinesis has increased quite a lot:
https://www.pombase.org/term_genes/GO:0000281
http://dev.pombase.kmr.nz/term_genes/GO:0000281

protein-containing complex assembly has increased a bit:
https://www.pombase.org/term_genes/GO:0065003
http://dev.pombase.kmr.nz/term_genes/GO:0065003

ValWood · 2020-10-14T08:09:56Z

Yes, I compared the slims too and all classes are either identical or increased with the new code.

old
transcription, DNA-templated | GO:0006351 | 385
new
transcription, DNA-templated | GO:0006351 | 390

I'm very happy, thanks
@cmungall
@kimrutherford
@balhoff

kimrutherford · 2020-10-14T08:49:24Z

The improvements are now live on the main site.

ValWood mentioned this issue May 10, 2018

problematic ontology commits in "conjugation with cellular fusion branch" geneontology/go-ontology#15641

Closed

This was referenced May 30, 2018

Amigo numbers over-inflated, but no annotation present geneontology/go-ontology#15740

Closed

Q conjugation with cellular fusion, annotation drop pombase/curation#1901

Closed

ValWood added the waiting waiting for external change (e.g. GO fix) label May 31, 2018

kimrutherford mentioned this issue Jun 5, 2018

Unstable output for owltools --save-closure-for-chado owlcollab/owltools#256

Open

mah11 mentioned this issue Jan 28, 2019

Handle output of new owltools version when computing closure #706

Closed

ValWood mentioned this issue Apr 18, 2019

flip flopping of annotation numbers geneontology/go-ontology#17171

Closed

ValWood mentioned this issue Sep 3, 2019

changes in transcription annotations #722

Closed

kimrutherford mentioned this issue Oct 14, 2020

big image issue pombase/website#1621

Closed

kimrutherford closed this as completed Oct 14, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ongoing flip-flopping of annotation #678

ongoing flip-flopping of annotation #678

ValWood commented May 9, 2018 •

edited

Loading

ValWood commented May 9, 2018

ValWood commented May 9, 2018

mah11 commented May 9, 2018

kimrutherford commented May 9, 2018

mah11 commented May 9, 2018

ValWood commented May 9, 2018

kimrutherford commented May 10, 2018

mah11 commented May 10, 2018

ValWood commented May 10, 2018

ValWood commented May 14, 2018 •

edited

Loading

mah11 commented May 14, 2018

ValWood commented May 14, 2018

cmungall commented May 14, 2018

kimrutherford commented May 14, 2018

ValWood commented May 15, 2018

mah11 commented May 15, 2018

ValWood commented May 16, 2018

ValWood commented May 16, 2018

ValWood commented May 31, 2018

kimrutherford commented Jun 5, 2018

ValWood commented Jun 5, 2018

ValWood commented Apr 18, 2019

kimrutherford commented Oct 14, 2020

ValWood commented Oct 14, 2020

kimrutherford commented Oct 14, 2020

ongoing flip-flopping of annotation #678

ongoing flip-flopping of annotation #678

Comments

ValWood commented May 9, 2018 • edited Loading

ValWood commented May 9, 2018

ValWood commented May 9, 2018

mah11 commented May 9, 2018

kimrutherford commented May 9, 2018

mah11 commented May 9, 2018

ValWood commented May 9, 2018

kimrutherford commented May 10, 2018

mah11 commented May 10, 2018

ValWood commented May 10, 2018

ValWood commented May 14, 2018 • edited Loading

mah11 commented May 14, 2018

ValWood commented May 14, 2018

cmungall commented May 14, 2018

kimrutherford commented May 14, 2018

ValWood commented May 15, 2018

mah11 commented May 15, 2018

ValWood commented May 16, 2018

ValWood commented May 16, 2018

ValWood commented May 31, 2018

kimrutherford commented Jun 5, 2018

ValWood commented Jun 5, 2018

ValWood commented Apr 18, 2019

kimrutherford commented Oct 14, 2020

ValWood commented Oct 14, 2020

kimrutherford commented Oct 14, 2020

ValWood commented May 9, 2018 •

edited

Loading

ValWood commented May 14, 2018 •

edited

Loading