Closed krchristie closed 1 year ago
Hi @krchristie and GO editors,
The term GO:0031427 - response to methotrexate (P) is imported in the Experimental Factor Ontology (EFO) and used by curators at the Genome-Wide Association Studies (GWAS) catalog. EFO imports many 'response to drug' terms, as you may see expanding the node here. As GO editors we addressed this issue 5 years ago in this ticket. In summary, at the time we resolved not to obsolete such terms, but rather place them in the do_not_manually_annotate subset, and add a definition comment "Note that this term is in the subset of terms that should not be used for direct manual annotation of gene products. It was created to be used for cross-referencing by other ontologies. Direct annotations to this term may be amended during annotation QC.". I do appreciate that 'response to methotrexate' should not be used to annotate when it's used as a proxy for folic acid, but the intent of GWAS curators is different (see here for the use of the term in the GWAS catralog). Would you please consider to keep GO:0031427, place it in the do_not_manually_annotate subset, and add a definition comment as above? (Note that no replacements are suggested in the annotation spreadsheet.)
I'm tagging my EFO colleagues @zoependlington and @kallia-p :
Thanks, Paola
@paolaroncaglia @dosumis @zoependlington Thank you for highlighting this Paola. In EFO we are currently reviewing the Response to terms (all GO 'response to stimulus' children and GO 'response to drug' subclasses). Historically EFO requested that terms were generated in GO and imported in EFO as per the EFO-GWAS catalog use cases, as Paola has explained. We are currently working on aligning Response to terms between EFO, GO and OBA. We plan to add all EFO 'response to' terms that do not have a GO id in OBA. Is it possible to delay obsoletion of GO:0031427 - response to methotrexate (P) until we have generated this term in OBA so as not to cause problems for our GWAS catalog users? If yes, I will make a note to come back to this ticket so that this term can be safely obsoleted without causing much disruption. If there are other 'response to drug' terms that GO is planning to obsolete we would appreciate if we are notified so that we can replace them.
Hi @paolaroncaglia @kallia-p @zoependlington @dosumis
We can definitely delay obsoletion until a plan is in place to make sure that GWAS curators are not disrupted.
Thanks @paolaroncaglia for pointing out the old ticket on Response to compounds used as assays #12228 . That is very helpful. Keeping these terms in GO certainly made sense 5 years ago. At this point in time though, if EFO can accomodate these terms, I think that would be a better place for them than GO, and better than keeping a somewhat haphazard set of these kinds of terms in GO marked in the do_not_manually_annotate subset.
Another specific term that has come up recently is "response to paclitaxel", see the response to paclitaxel #21750 ticket.
GO is also planning to merge the term "response to drug" into the term "response to "xenobiotic", see this ticket: handling terms for xenobiotic role and removing drug role #19460
As GO is moving towards removing these kinds of terms, and EFO is now a good place for them, it seems like it might be a good time to come up with a plan to move forwards with EFO taking over terms like this that really don't belong in GO.
@pgaudet what do you think about this idea? Maybe at some point, it would be good to see if EFO curators can attend our GO editors meeting (currently Mondays at 9 am Pacific, 5 pm UK), if that time would work for you.
Hi @krchristie ,
Thank you very much for your reply. @kallia-p has outlined a detailed plan here. Based on that, I think that there may be no need for an urgent EFO-GO meeting, but I'll leave this to Kallia, as I'm not involved in GWAS work at the moment.
Wishing you a great day, Paola
Hi @krchristie @paolaroncaglia @zoependlington Thank you for informing us of your proposed changes and for explaining the rationale behind these changes. EFO users such as the GWAS catalog and Open Targets would prefer to cross-reference their pharmacogenomic response terms to GO biological processes and therefore any changes to the broad parental classes such as the merge of 'response to drug' with the 'response to xenobiotic' are more important to our users than the children of this branch for example, response to paclitaxel. Since @zoependlington is the contact person for OTAR, I will liaise with her to discuss ways of tackling these proposed changes. Thanks, Kallia
Hi @krchristie @zoependlington @paolaroncaglia
I think, given the timing it would be best if we could both approach the 'response to' terms in a systematic way rather than as part of sporadic tickets about specific drugs. If you agree we could start discussing this in a new ticket. If needed I can join one of your future meetings to discuss and plan around the proposed work with regards to the alignment between GO, EFO and OBA 'response to' terms. Please let me know which of the two you prefer and if you think a meeting is warranted which forthcoming Monday we can allocate to this task.
I provide a bit of background and context to the preliminary work that I have done thus far. I am currently reviewing all of the terms that are in GO and have been imported in EFO (N=75 GO terms, see table below which comprises the EFO created 'response to' specific drug terms as well). I am also examining the OBA cross referenced terms with GO and the overlap between the three ontologies. With regards to EFO some of the imported GO terms are high level grouping terms with the main one being 'response to stimulus' of which 'response to chemical' is a Subclass Of. 'Response to drug' is a Subclass Of 'response to chemical' and I have made a note that you plan to merge this with 'response to xenobiotic'. Further down the GO hierarchy i.e. in the 'response to drug' branch there is some structure incorporated in GO that several EFO terms sit under for example the GO 'response to antipsychotic drug' and GO 'response to antineoplastic agent' have a number of children with either GO or EFO IDs for specific antipsychotic drugs and cancer therapy agents (with one of them being the GO 'response to paclitaxel' which you are planning to obsolete). Small changes like these will have less of an effect to our users but changes in higher level grouping terms will cause much disruption and I wonder why do you think EFO would be a better place for them than GO, and what is the hazard of keeping these kind of terms in GO.
I reviewed the tickets in trying to understand how GO views usage of the 'response to' terms in general. In my view an organism's response to an exogenous stimulus whether this is food intake, stress or a pharmacologic intervention constitutes a biological process; variations in our our genes interact with our environment. Many of the GWAS catalog studies associate variants with phenotypes that measure responses. Pharmacogenomics studies will specifically look for variants that modulate efficacy of a drug or adverse/toxic reactions. The GWAS catalog contains associations from gene x environment interactions studies as well. I think it also useful for researchers to know that (a variant affecting expression or function of) gene X which is associated with an adverse reaction to a drug Y (GO 'response to term') is linked to GO biological process Z.
Thanks!
Hi @kallia-p @zoependlington @paolaroncaglia
I think, given the timing it would be best if we could both approach the 'response to' terms in a systematic way rather than as part of sporadic tickets about specific drugs. If you agree we could start discussing this in a new ticket. If needed I can join one of your future meetings to discuss and plan around the proposed work with regards to the alignment between GO, EFO and OBA 'response to' terms. Please let me know which of the two you prefer and if you think a meeting is warranted which forthcoming Monday we can allocate to this task.
I think a systematic approach would be good. I'll bring this up at the editors meeting next Monday, just as a preliminary discussion to check if other agree with me about a systematic approach at this time.
I provide a bit of background and context to the preliminary work that I have done thus far. I am currently reviewing all of the terms that are in GO and have been imported in EFO (N=75 GO terms, see table below which comprises the EFO created 'response to' specific drug terms as well). I am also examining the OBA cross referenced terms with GO and the overlap between the three ontologies. With regards to EFO some of the imported GO terms are high level grouping terms with the main one being 'response to stimulus' of which 'response to chemical' is a Subclass Of. 'Response to drug' is a Subclass Of 'response to chemical' and I have made a note that you plan to merge this with 'response to xenobiotic'. Further down the GO hierarchy i.e. in the 'response to drug' branch there is some structure incorporated in GO that several EFO terms sit under for example the GO 'response to antipsychotic drug' and GO 'response to antineoplastic agent' have a number of children with either GO or EFO IDs for specific antipsychotic drugs and cancer therapy agents (with one of them being the GO 'response to paclitaxel' which you are planning to obsolete). Small changes like these will have less of an effect to our users but changes in higher level grouping terms will cause much disruption and I wonder why do you think EFO would be a better place for them than GO, and what is the hazard of keeping these kind of terms in GO.
I reviewed the tickets in trying to understand how GO views usage of the 'response to' terms in general. In my view an organism's response to an exogenous stimulus whether this is food intake, stress or a pharmacologic intervention constitutes a biological process; variations in our our genes interact with our environment. Many of the GWAS catalog studies associate variants with phenotypes that measure responses. Pharmacogenomics studies will specifically look for variants that modulate efficacy of a drug or adverse/toxic reactions. The GWAS catalog contains associations from gene x environment interactions studies as well. I think it also useful for researchers to know that (a variant affecting expression or function of) gene X which is associated with an adverse reaction to a drug Y (GO 'response to term') is linked to GO biological process Z.
We completely agree that it is useful to know this type of information; we just don't think all of it belongs in GO or should be captured in GO annotations.
There are a couple principles guiding our plans. One is that GO's purpose & scope is to make statements about the normal purpose of gene products. Terms invoking compounds that are only synthetic compounds don't reflect the normal purpose of a gene.
Another issue is that concepts represented by ChEBI roles like "drug", "vitamin" are context specific, depending on things like the specific usage for a compound that may be a drug in some contexts but not others, or the species for a compound that is a vitamin in some species but not others. For many of these ChEBI roles, we feel that the role is grouping things in ways that are not always true, which is against the principles of an ontology. For "drug" specifically, this is intrinsically out of scope for GO, though we do completely agree that "response to xenobiotic" is a normal process for organisms to deal with their environment. However, within an ontology like GO that encompasses many species, we can not specifiy whether any specific compound is a xenobiotic or not, as it depends on which species you are considering.
I hope that helps explain our thinking, but a more full explanation would probably be best on a call, either a full editors call, or we could set up something smaller on a more flexible schedule.
I look forward to working with you on this.
-Karen
@krchristie Thank you for your reply, looking forward to an update when you discuss this with other editors. Kallia
Hi @kallia-p @zoependlington @paolaroncaglia
We discussed this at the GO Editors call on 9/27/21. Here's a quick summary:
I've made another ticket to deal with this project, so I'll put all further comments about this project in the new ticket.
Thanks
Karen
3 EXP to cellular response to methotrexate by GeneDB
The proposal has been made to obsolete all the methotrexate terms for the reason given below.
This was discussed at the GO annotation call on 9/7/2021.
GO term ID and Label
GO:0051870 - methotrexate binding (F)
GO:0015350 - methotrexate transmembrane transporter activity (F)
GO:0051958 - methotrexate transport (P)
GO:0031427 - response to methotrexate (P)
Reason for deprecation The reason for obsoletion is that methotrexate is a synthetic compound that is an analog of folic acid. It's effects on organisms are usually due to its similarity to folic acid, not due to any organism being evolved to deal with methotrexate specifically.
"Replace by" term (ID and label) If all annotations can safely be moved to that term
NA
"Consider" term(s) (ID and label) Suggestions for reannotation
GO:0051870 - methotrexate binding (F) => consider: -- GO:0005542 - folic acid binding -- DELETING annotation
GO:0015350 - methotrexate transmembrane transporter activity (F) => consider: -- GO:0008517 - folic acid transmembrane transporter -- GO:0015347 - sodium-independent organic anion transmembrane transporter activity (make new Na-ind polyspecific term?? or make this term polyspecific??) -- DELETING annotation
GO:0051958 - methotrexate transport (P) => consider: -- GO:0015884 - folic acid transport -- DELETING annotation
GO:0031427 - response to methotrexate (P) => consider: -- GO:new - negative regulation of dihydrofolate reductase activity -- DELETING annotation
Are there annotations to this term?
How many EXP: 32 experimental annotations by ARUK-UCL, BHF-UCL, CAFA, RGD, or UniProt
GeneDB & MGI only have annotations by ISS or ISO
Ensembl, GOC, and ZFIN only have annotations by IEA
no
no