search

geneontology / go-ontology

Source ontology files for the Gene Ontology
http://geneontology.org/page/download-ontology
Creative Commons Attribution 4.0 International
219 stars 40 forks source link

Interferon terms #14140

Closed Ayllonbe closed 6 years ago

Ayllonbe commented 7 years ago

Good morning, I would like to kno w why there is not an intermediary term between cytokine production and the different type of interferon production. I suggest the term interferon production between them. And to do the same in every cytokine related terms and the different types of tinterferon. What do you think about that?

pgaudet commented 6 years ago

Hello @Ayllonbe

Sorry about the long delay ! This is what the GO has today:

What are you suggesting we add ?

Thanks, Pascale

Ayllonbe commented 6 years ago

Hi!

I think that adding the terms interleukin production and interferon production as subclass of cytokine is better than using a narrow synonym. I would propose that:

cytokine production

Thanks

Aaron

pgaudet commented 6 years ago

Hi,

I've added the two parents.

image
Ayllonbe commented 6 years ago

Hi!

Yes, I think that it is more evident to explore the GO structure. If the change you have in every cytokine case (cytokine production, cytokine response....) could make the ontology more readable. Only if the community is agreeing with that.

thank you for listening to me.

Regards

Aaron Ayllon Benitez

On 13 February 2018 at 13:38, pgaudet notifications@github.com wrote:

Hi,

I've added the two parents.

[image: image] https://user-images.githubusercontent.com/4782928/36150207-320507d6-10c3-11e8-85e0-db6c59880ae7.png

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/14140#issuecomment-365254853, or mute the thread https://github.com/notifications/unsubscribe-auth/AG79o-DJ9H8JTt1fPMan4T9Li2zRy30cks5tUYJRgaJpZM4PJyo2 .

pgaudet commented 6 years ago

@addiehl @RLovering (who else works in immune-related processes?) Is that OK with you ?

RLovering commented 6 years ago

this makes sense to me

addiehl commented 6 years ago

I am totally opposed to this change, as I have argued on multiple occasions in the deep past. Unfortunately, it is difficult to dredge up some of my earlier argumentation as it is buried in dead email accounts and old SourceForge entries, so I will have to try to summarize it here.

The division of cytokines into "interleukins" and "non-interleukins" is an arbitrary lexical division that reflects the tortured history of naming proteins in immunology rather than any shared structural or functional characteristics among the proteins called interleukins vs those that are not. This is in sharp contrast to the chemokine (and their receptors), which share strong structural homology and general functional roles, and similarly to the interferons, which again share strong structural homology and somewhat more varied but general functional roles.

I suggest that all parties to this discussion read PMID:12103349 and PMID:12464567. The designation of a newly discovered cytokine as an interleukin was at one point a point of pride and envy among immunologists and this trend was recognized fairly early on (PMID:1934243), but the application of the "interleukin" designation was arbitrary in that some researchers just didn't bother. On the other hand, we have such things as the non-existant "IL-14", which was claimed on fairly weak evidence and genomic sequencing data that was disavowed by most of the original authors (PMID:8755619), reflected in the fact that there is no official IL14 gene in human or mouse.

The argument that adding interleukin grouping terms makes the ontology more "readable" is a completely non-ontological notion. When we create grouping terms in the GO, such terms should be created because of commonalities that go beyond arbitrary naming. There are much more useful ways one could group cytokine related terms, if people actually concerned themselves with structure and biological activities, and indeed, I had once worked on this at one point over a decade ago but gave up due to other responsibilities. Key point: If you are searching for terms by reading the ontology instead of using a search function, then you need better training in the use of AmiGO and other ontology browsers.

Sorry for being so contrary, as always, Alex

Ayllonbe commented 6 years ago

Hi @addiehl https://github.com/addiehl,

Thank for your honest and well-documented response. My purpose with that extension in the cytokines classification was totally from the point of view of the application. I work with similarity method exploiting the bio-ontologies and I found weird that interferon-alpha production has the same similarity with interferon-gamma production and interleukin IL-2 production only because they share the same ancestors subClassOf. I am not a specialist in the cytokines fields (I have only some notions), and obviously, IFN and IL have strong structural homology and strong general functions, but they have a difference between them (still the strong similarity), for that, they have different names (chemokine, interferon, interleukine) and not only interleukine. Well, maybe I am totally wrong with this suggestion, but I thought that could be a good solution to similarity constraints.

Again, thank you for your response

Regards

Aaron

On 13 February 2018 at 16:25, Alexander Diehl notifications@github.com wrote:

I am totally opposed to this change, as I have argued on multiple occasions in the deep past. Unfortunately, it is difficult to dredge up some of my earlier argumentation as it is buried in dead email accounts and old SourceForge entries, so I will have to try to summarize it here.

The division of cytokines into "interleukins" and "non-interleukins" is an arbitrary lexical division that reflects the tortured history of naming proteins in immunology rather than any shared structural or functional characteristics among the proteins called interleukins vs those that are not. This is in sharp contrast to the chemokine (and their receptors), which share strong structural homology and general functional roles, and similarly to the interferons, which again share strong structural homology and somewhat more varied but general functional roles.

I suggest that all parties to this discussion read PMID:12103349 and PMID:12464567. The designation of a newly discovered cytokine as an interleukin was at one point a point of pride and envy among immunologists and this trend was recognized fairly early on (PMID:1934243), but the application of the "interleukin" designation was arbitrary in that some researchers just didn't bother. On the other hand, we have such things as the non-existant "IL-14", which was claimed on fairly weak evidence and genomic sequencing data that was disavowed by most of the original authors (PMID:8755619), reflected in the fact that there is no official IL14 gene in human or mouse.

The argument that adding interleukin grouping terms makes the ontology more "readable" is a completely non-ontological notion. When we create grouping terms in the GO, such terms should be created because of commonalities that go beyond arbitrary naming. There are much more useful ways one could group cytokine related terms, if people actually concerned themselves with structure and biological activities, and indeed, I had once worked on this at one point over a decade ago but gave up due to other responsibilities. Key point: If you are searching for terms by reading the ontology instead of using a search function, then you need better training in the use of AmiGO and other ontology browsers.

Sorry for being so contrary, as always, Alex

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/14140#issuecomment-365299880, or mute the thread https://github.com/notifications/unsubscribe-auth/AG79o7F4-27MWFIMSMD497mygkl1sCFBks5tUalYgaJpZM4PJyo2 .

addiehl commented 6 years ago

The only shared similarity across all the cytokines that happen to be called interleukins is the name. This is not the situation for the chemokines and the interferons. There are a number of protein families represented by certain subsets of interleukins, but some of those protein families have members which do not happen to be named interleukin-X, even though in a different universe they might well be.

Grouping terms based on the "interleukin" name only serve to suggest that naming, no matter how arbitrary, is an appropriate principle for structuring the Gene Ontology. This is in fact completely contrary to why the GO was created in the first place.

Ayllonbe commented 6 years ago

So,

What do you think about to create only an intermediary class interferon between cytokine and all type of interferons?

On 13 February 2018 at 18:18, Alexander Diehl notifications@github.com wrote:

The only shared similarity across all the cytokines that happen to be called interleukins is the name. This is not the situation for the chemokines and the interferons. There are a number of protein families represented by certain subsets of interleukins, but some of those protein families have members which do not happen to be named interleukin-X, even though in a different universe they might well be.

Grouping terms based on the "interleukin" name only serve to suggest that naming, no matter how arbitrary, is an appropriate principle for structuring the Gene Ontology. This is in fact completely contrary to why the GO was created in the first place.

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/14140#issuecomment-365337851, or mute the thread https://github.com/notifications/unsubscribe-auth/AG79o---vGMoQLfqY7rf0NMbSNKjkVnFks5tUcPqgaJpZM4PJyo2 .

addiehl commented 6 years ago

Interferons are grouped by immunologists into three types, based on structural relatedness and functional characteristics. Type I interferons include the interferon-alpha and interferon-beta as well as some lesser known interferons, and were originally associated with anti-viral immunity, though they have other immune modulatory effects. Type II interferon, AKA interferon-gamma, is very well studied cytokine with a wide variety of effects on various cell types and is strongly associated with Type I immune responses. Type III interferons, interferon-lamda 1, 2, and 3, are a separate group that are more recently discovered and associated with antiviral responses. Interferon-lamda 1, 2, and 3 are also known as IL29, IL28A and IL28B, respectively, which points again to the arbitrariness of the application of the interleukin designation.

The grouping of the interferons by type is based again on structural and function similarities. I do not strongly support a common "interferon production" term per se (as well as analogous terms), as we already have separate grouping terms for Type I and Type III interferon production as well as an interferon-gamma production term, but given the loose structural and functional similarities among the interferons, I can accept such a class, because of the structural relatedness and association with anti-viral immune responses, among other processes, seen for all interferons.

cmungall commented 6 years ago

Alex and all,

How can we get better at having institutional memory about these things. As you say there is a lot of info in trackers but hard to find.

What I often do in other ontologies is create a class and immediately obsolete it, e.g. for "interleukin production". We can then provide a short textual description and a link to the wiki (which can itself have links to tickets if need be).

We also have 'interleukin production' as a narrow syn. I think this is appropriate. But we can also annotate this with a link to a ticket or page discussing why a term hasn't been created. Currently there is not a good way to know whether NARROW means (a) this is a true subclass but not we haven't got round to making a subclass or it's not important enough (b) this is an inappropriate concept but technically the scope is still narrower.

We can also get better at using the wiki. We could have a category that is 'biological modeling' or 'ontology content' and tag pages describing design decisions in the ontology. This would be separate from the technical editors docs on RTD. It may interweave with the DPs, but in this case it's different. Yes this is work, but going back to find previous discussions is also work...

addiehl commented 6 years ago

Hi Chris,

I agree with your suggestions and hope that some version of them can be implemented.

Also, I did find a 2014 version of my objections, and I know there are email threads as well: https://sourceforge.net/p/geneontology/ontology-requests/11250/

-- Alex

pgaudet commented 6 years ago

HI @cmungall Is it possible to add a db cross reference to the GO wiki for the interleukin/interferon synonym? Otherwise I can add it to the 'comment' for now.

Thanks, Pascale

cmungall commented 6 years ago

I think that's a good idea. This would be done as an annotation axiom using the has dbxref property

On Feb 22, 2018 7:13 AM, "pgaudet" notifications@github.com wrote:

HI @cmungall https://github.com/cmungall Is it possible to add a db cross reference to the GO wiki for the interleukin/interferon synonym? Otherwise I can add it to the 'comment' for now.

Thanks, Pascale

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/14140#issuecomment-367711854, or mute the thread https://github.com/notifications/unsubscribe-auth/AADGOf0F6b6KnQ6VGXg7uJQKtaEX3tHkks5tXYQhgaJpZM4PJyo2 .

pgaudet commented 6 years ago

Hello @Ayllonbe So I ended up adding a link to this page: http://wiki.geneontology.org/index.php/Why_isn%27t_interleukin_in_GO%3F

to the term 'cytokine production'. Does that address the issue ? (Maybe there are other terms to which this should be added - although there are so many of these terms I dont think we can be exhaustive, at least not with some manual edits).

WRT interferon, I don't think it's worth it to create a grouping class for just two terms.

Let me know if this addresses the point you were raising, and apologies for the long delay for dealing with this !

Thanks, Pascale

pgaudet commented 6 years ago

I have added the same x-ref to the interleukin synonyms of:

I didn't do

Since these didn't have any 'interleukin' synonym. Let me know if I should add these.

Thanks, Pascale

Ayllonbe commented 6 years ago

Hello Pascale,

Yes the issue is about the term "cytokine production". I think the addition to the interleukin synonyms is great and it is not necessary to add interleukine.

Thanks for listening and don't worry for the delay.

Regards

On 23 May 2018 at 13:12, pgaudet notifications@github.com wrote:

I have added the same x-ref to the interleukin synonyms of:

  • GO:0004896 cytokine receptor activity
  • GO:0050663 cytokine secretion

I didn't do

  • response to cytokine
  • cytokine-mediated signaling pathway
  • regulation of cytokine production

Since these didn't have any 'interleukin' synonym. Let me know if I should add these.

Thanks, Pascale

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/14140#issuecomment-391309594, or mute the thread https://github.com/notifications/unsubscribe-auth/AG79ow1AtqEJTNc7lHuOz1ESJFbN_lU0ks5t1UQLgaJpZM4PJyo2 .

-- Aarón Ayllón Benítez*PhD Student in Bioinformatics*INSERM U1219 - Erias team and CNRS (UMR 5800) - MABioVis team - LaBRI, Université Bordeauxp:+33 (0)5 40 00 38 83a:351, cours de la Libération F-33405 Talence cedex, France e:ayllonbenitez.aaron@gmail.com

pgaudet commented 6 years ago

Great ! Thanks for your feedback.