Add entity type 'protein' for IKR evidence

[pgaudet]

We want to do this to allow using MEROPs (and perhaps other resources) as evidence for IKR annotations.

Specific example from @sjm41

In that example:
Entity: A1ZBU9 (CG13423/FBgn0034513)
Qualifier: NOT|enables
GO: GO:0008233 (peptidase activity)
Evidence: IKR
Reference: GO_REF:0000047
With/from: MEROPS:C01.UNB [<- this is the bit I’d like to add, but I can’t in P2GO]

That MEROPS ID corresponds to this non-peptidase family entry at MEROPS:
https://www.ebi.ac.uk/merops/cgi-bin/pepsum?mid=C01.UNB
https://www.ebi.ac.uk/merops/cgi-bin/sequence_data?mid=C01.UNB

So in summary, I’d like to make a ’NOT peptidase activity’ annotation to proteins that have peptidase domains but are classified as ’non-peptidase homologs’ at MEROPS because they lack key catalytic residues.

PR is here:#1881

Thanks, Pascale

[sjm41]

Thanks @pgaudet . Sounds like this would fix things for my use case. I think the alternative is to change the "type_name" of the MEROPS entry in the GO xrefs metadata to "gene/protein family" - see #1875. I'm not sure which solution is better/easier??

FYI: Looking at all MF annotations in QuickGO, I don't see any annotations using 'MEROPS' in the with/from column, so any change here won't affect any existing annotations - it will just determine how we can use MEROPS IDs in IKR annotations going forward.

[pgaudet]

Looking at all MF annotations in QuickGO, I don't see any annotations using 'MEROPS' in the with/from column, so any change here won't affect any existing annotations - it will just determine how we can use MEROPS IDs in IKR annotations going forward.

Right! since it wasn't allowed until now. Thanks for checking!

[sjm41]

Right, but given there's a MEROPS entry in the GO xrefs metadata file (#1875) and I don't know how long that's been there, I wanted to check if there's been any usage of MEROPS IDs in any context (e.g. ISS?) to date. Seems not.

[pgaudet]

Hi @sjm41
This is working using the syntax provided by identifiers.org, see https://registry.identifiers.org/registry/merops#

which is something like C01.### (where ### is a number)

Do you need the identifiers with 3 letters after the dot? We cannot find an 'official source' for that format, but we can relax the rules if needed.

Thanks, Pascale

[sjm41]

Hi @pgaudet
Thanks for following up. Yes, the ### bit should be relaxed to be any three letters or numbers. It's true that the MEROPS ID is defined [here](https://www.ebi.ac.uk/merops/about/glossary.shtml#MeropsID) with the ### bit being 3 numbers, but as the example at the top of this page shows (C01.UNB), that's not always the case in practice.....

[sjm41]

Looking at the MEROPS IDs I want to use (at least), the ### bit corresponds to one of 3 regular expressions:

• 3 numbers (e.g. 984)
• 3 letters (e.g. UPA)
• 1 letter followed by two numbers (e.g. A65)

[sjm41]

This is working using the syntax provided by identifiers.org
I just checked in Protein2GO and I don't (yet) see MEROPS as an option in the 'with' drop-down menu when using IKR (the only option is still 'InterPro').

I guess this may be a separate/downstream issue, maybe specific to P2GO??
I tried reviewing the email thread "GO_REF attribution for internal NOT annotations (IKR)", but I'm still unclear what needs to be done for P2GO to allow a MEROPS ID in this use case. Do we need another chat with Alex?

[pgaudet]

@alexsign and I edited the regular expression again, see
#1908

Let's hope this time it works!

[sjm41]

Thanks @pgaudet and @alexsign for continuing to work on this!

Good news is that I now see 'MEROPS' (and 'MEROPS_fam') in the drop-down menu on protein2GO:

Other good news is that a ID like "S09.986" (ie. three numbers after the dot) or "S01.A18" (one letter + two numbers after the dot) is a valid entry (entered with 'MEROPS' selected in the drop-down).

Bad news is that IDs like "C85.UNA" (ie. three letters after the dot) isn't allowed. Can you check the regex again? Many thanks!!

[sjm41]

Can we also clarify the difference between 'MEROPS' and 'MEROPS_fam' in the menu?

[pgaudet]

Should now work

[sjm41]

Yep, confirming that IDs like "C85.UNA" (ie. three letters after the dot) now work. Thanks!

[sjm41]

Notes from quick chat with Pascale:

• as we know, the "MEROPS" field is for the "MEROPS ID" - XXX.YYY for a specific peptidase
• the 'MEROPS_fam' field seems to be for a family ID, i.e. just the XXX bit (e.g. S01 or C85), but a MEROPS family contains many different peptidases, including both active and inactive, so we can't see how/why a curator could use a family ID in a GO annotation. Pascale will look into removing this option.
• we need to add to the documentation to explain the current, valid entry types in the 'with/from' field with example entries

[pgaudet]

Remove protein family for IKR

https://github.com/geneontology/go-site/compare/master...pgaudet-patch-71?quick_pull=1

Hopefully this should do it.