regulates_o relations

[dosumis]

These have been extensively used by MGI. Their meaning has until recently been obscure to most curators, but can be made much clearer by some simple renaming.

e.g.

regulates_o_development_of -> regulates development of
usage: Use this relation to link a process that regulates a developmental process to a developing structure.

Unfortunately, 'regulates_o_development_of' will remain as an ID and as how the relationship is stored in GAF files.

I've added new labels and usage statements to most of these.

Decision to make: should we open up usage for all annotators?

Note - there are some challenges in making the labels visible in tools. Can be done, but will take some time and we need to decide where it is appropriate to display label and where ID.

[dosumis]

@ukemi, @RLovering - comments please

[cmungall]

Why complicate things? I think if all annotators get the same training, then the lexical pattern can be learned in 5 minutes. Ideally all annotators would also look at the OWL definition, not sure how easy that is in p2go.

Note that these are all auto-expanded in amigo, so the user doesn't see them. Maybe p2go could use the same expansion as in amigo?

[RLovering]

Hi Chris

Please could you be more specific. I don't understand what you think is the complication, the previous email I was reading had a lot in it, are you saying the whole thing is wrong?

What instructions will you give the annotators?

The range and domain should be what drives the use.

What is the lexical pattern, please could you teach this to me!!

I don't think you appreciate the breadth of misuse that exists in GO now, I was looking at David's email and I have no idea why someone would create GO:0022409 positive regulation of cell-cell adhesion regulates_o_has_participant(CL:0000031). I am not saying this is wrong it is just that I think I would have created GO:0022409 positive regulation of cell-cell adhesion occurs_in (CL:0000031)

I do see that the cell is participating in the adhesion but at the same time use of regulates_o_has_participant would not be my first choice. But there again there was no documentation for any of the regulates_o_ relationships and we were told these might be removed so I haven't been paying them much attention.

Wrt OWL which of the above would create positive regulation of neuroblast cell-cell adhesion and how would you use C16 to create:
positive regulation of neuroblast cell- epithelial cell adhesion?

Plus please explain what you mean by auto-expanded in amigo. It just looks to me like the C16 information is just displayed as it is added (see screenshot)

Thanks

Ruth

From: Chris Mungall <[email protected]mailto:[email protected]>
Reply-To: geneontology/annotation_extensions <[email protected]mailto:[email protected]>
Date: Thursday, 11 June 2015 15:55
To: geneontology/annotation_extensions <[email protected]mailto:[email protected]>
Cc: Ruth Lovering <[email protected]mailto:[email protected]>
Subject: Re: [annotation_extensions] regulates_o relations (#9)

Why complicate things? I think if all annotators get the same training, then the lexical pattern can be learned in 5 minutes. Ideally all annotators would also look at the OWL definition, not sure how easy that is in p2go.

Note that these are all auto-expanded in amigo, so the user doesn't see them. Maybe p2go could use the same expansion as in amigo?

Reply to this email directly or view it on GitHubhttps://github.com//issues/9#issuecomment-111162224.

[cmungall]

Did you mean to attach a screenshot?

Here is the amigo view of

There's a more formal description of this in the methods section of the Huntley et al paper

The lexical pattern is

foo_o_bar means a foo followed by a bar

[ukemi]

I think it would be great if people started to use the regulation chains. I think it is a great step to expand on the has_regulation_target relation to try to pin it down. One concern is about the expression term. Different groups annotate different entities, so if we want to go down that route I think we have to ask 'expression of what'?. We (MGI) put in a gene ID, so I think that is ok but maybe better to go all the way down the route of regulation_o_has_input, regulation_o_has_output, regulates_o_gene_product_of etc. I think using these will be more explicit and better in the long run, but I think the gene product relations need to be fleshed out a bit. I bet at some point, people are going to want to start using splice variants, alternate transcripts, protein isoforms etc and then we will have to deal with that.
The original idea was to restrict relations to terms and then provide examples of how the relations are used for those terms. We probably have precomposed examples in the ontology that we could already use for almost all of them. I don't know if that has gone by the wayside.

[dosumis]

regulates_o_gene_product_of etc. I think using these will be more explicit and better in the long run, but I think the gene product relations need to be fleshed out a bit.

We now have

• regualtes expression pf
  • regulates translation of
  • regulates transcription of

All of these have range: gene.

This allows for the relation heirarchy => inference from 'regulates transcriptiontranslation of' -> 'regulates expression of'. It doesn't allow for specifying splice varients, alternate transcripts, protein isoforms. It would be challenging to allow these and keep this inference. It might be better to just use 'regulates levels of' for specific products.