[Annotation] [Refgenome] Process annotation for ribosomal proteins
Harold Drabkin
hjd at informatics.jax.org
Fri Aug 29 14:31:16 PDT 2008
Rama Balakrishnan wrote:
> Considering that ribosomal genes are the Ref.Genome targets for Aug, I
> would like to revive this discussion.
>
> There was no clear answer on what evidence code is best for PROCESS
> annotation in a situation where all you know is that the protein is a
> subunit of a larger well established complex?
If for example, there is a good experimental to a complex like a
ribosome, one could use an IC to translation based on the GO id for the
ribosome and it's reference., I would think. Unfortunately for mouse,
the ribosomal protein genes don't even have a real experiment to show
they are in the ribosome 8-(
Most cloned via sequence similarity etc. etc.
hjd
>
> Here is Karen's summary of the whole issue.
>>
>> We don't have a problem with the complex annotation. It's fine to say
>> that protein/gene X is in a complex by IDA.
>>
>> Our issue is with the process annotations for things where the only
>> thing you know is that it is part of a complex, e.g. RPS25A is part
>> of the ribosome, or UTP21 is part of the 90S preribosome. For both of
>> these proteins, the sum total of the available experimental evidence
>> is that they are part of the specified complex. There is no
>> individual biochemical or genetic characterization. Nevertheless, the
>> researchers who work on these complexes feel that being part of a big
>> complex, whose function is known, e.g. translation for the ribosome
>> and ribosome biogenesis and assembly for the 90S preribosome, is good
>> evidence that a given gene product participates in that process.
>>
>> Before the 2006 Annotation Camp decision that IPI could only be used
>> when you know it is direct, we used to use IPI for the process
>> annotations. Now, I'm not really sure what evidence code to use in
>> order to be able to annotate UTP21 to the process of "ribosome
>> biogenesis and assembly". Since I can't put anything specific in the
>> with field because I only know that UTP21 is part of the complex, but
>> not whether it interacted specifically with any of the tagged
>> proteins used to pull it down, we don't use IPI anymore. The
>> experiment showed that UTP21 came down in the 90S preribosome
>> complex, but while this is IDA for the component annotation, is it
>> IDA for the process annotation?
>>
>> It has also been suggested to use IC from the GOID from the component
>> term. However, this hides the fact that there is an experimental
>> basis to believe that protein X is involved in a process Y. We're
>> really not keen on this idea.
>>
>> To me, it seems that IPI is really the best explanation of the type
>> of evidence for the process annotation, i.e. we know that protein X
>> physically interacts with these other gene products as part of
>> complex Z. I think we should either go back to the pre-2006 Annot
>> Camp guidelines for IPI, which did NOT require direct interaction, or
>> allow some form of complex ID in the with field to cover these
>> situations.
>>
>> -Karen
>
> I would greatly appreciate your feedback.
>
> Thanks,
>
> Rama
>
>
>
>
>
>
>
>
>
>
>
>>
>>> Judy
>>>> To use the 'co-localizes' qualifier to qualify the identifier in
>>>> the with/from column, which is basically supporting information for
>>>> the evidence code, would be a very different use of the qualifier.
>>>> We have previously decided that we did not want to mix and match
>>>> what these qualifiers meant because it would be confusing.
>>>> Considering that the ID put in the with column is supporting
>>>> evidence, I think that it would be OK to allow IPI with "GOID for a
>>>> complex" to include both where the annotated gene product is part
>>>> of the complex and where the annotated gene product interacts with
>>>> the complex. In the specific example brought up, we are talking
>>>> about making a process annotation, and I think that it would be
>>>> valid to make process annotations based either on a given gene
>>>> product being in a complex or interacting with it.
>>>> -Karen
>>>> On Fri, 27 Jun 2008, Judith Blake wrote:
>>>>> Here's what I was thinking.
>>>>> IPI for the binding
>>>>> with the GO:complexID in the "with" field
>>>>> and the 'co-localizes' as the qualifier.
>>>>> Thus, the gene product 'colocalizes' with 'complexID' by evidence
>>>>> code 'IPI'
>>>>> which is different than
>>>>> Rama Balakrishnan wrote:
>>>>>> Hi Judy,
>>>>>> Comment below.
>>>>>> On Jun 25, 2008, at 4:51 AM, Judith Blake wrote:
>>>>>>> Rama,
>>>>>>> I agree that the solution needs to go beyond ReactomeIDs.
>>>>>>> However, there are many other complex resources, and my summer
>>>>>>> intern is in the process of collecting a more global (one would
>>>>>>> hope comprehensive) list of complexes with IDs from these other
>>>>>>> resources. So putting a complex ID in the 'with' field seems
>>>>>>> correct. And allowing the GO:ID seems correct too
>>>>>>> I think distinguishing between 'member of complex' and
>>>>>>> 'interacts with complex' is very important. The CC assignment
>>>>>>> places a gp in a complex. I think using the IPI code here, while
>>>>>>> technically defensible, might result in the confusion. Would
>>>>>>> this be a place that IPI with 'co-localizes' would clarify the
>>>>>>> distinction?
>>>>>> Hmm...I don't think I follow your last sentence completely.
>>>>>> So far, we have used Qualifiers to qualify the GO term and not
>>>>>> the evidence. And we don't use qualifiers for Process terms.
>>>>>> Contributes_to is specifically meant for Function, and
>>>>>> co_localizes is meant for CC.
>>>>>> Thanks,
>>>>>> Rama
>>>>>>> judy
>>>>>>> Rama Balakrishnan wrote:
>>>>>>>> Hi all,
>>>>>>>> I am sure all the other groups annotating ribosomal genes are
>>>>>>>> facing the same issue. I would like to hear from you about the
>>>>>>>> new proposal for IPI or if you have other suggestions.
>>>>>>>> Thanks,
>>>>>>>> Rama
>>>>>>>> On Jun 18, 2008, at 2:53 PM, Rama Balakrishnan wrote:
>>>>>>>>> Hi,
>>>>>>>>> 1) Although there is a REACTOME ID for the yeast ribosome
>>>>>>>>> complex, I think a better solution would be to allow the GOID
>>>>>>>>> for the complex in the 'with' column for IPI. This is because,
>>>>>>>>> Reactome has IDs only for human and yeast, and this solution
>>>>>>>>> won't scale for other organisms.
>>>>>>>>> 2) If we move towards the idea of using IPI in this fashion,
>>>>>>>>> we also need to update the documentation for IPI because by
>>>>>>>>> saying that the subunit interacts physically with the complex,
>>>>>>>>> one could infer that the subunit peripherally interacts with
>>>>>>>>> the complex as oppose to inferring that the subunit is part of
>>>>>>>>> the complex.
>>>>>>>>> Thanks,
>>>>>>>>> Rama
>>>>>>>>> On Jun 18, 2008, at 9:25 AM, D'Eustachio, Peter wrote:
>>>>>>>>>> Hi, Midori,
>>>>>>>>>> We do have such identifiers (though in general - this case is
>>>>>>>>>> an exception - only for human proteins and complexes) and
>>>>>>>>>> this sounds like a legitimate use of them, but it also seems
>>>>>>>>>> like a kind of hack around the basic problem. Earlier in this
>>>>>>>>>> thread, someone pointed to one paper that supports the
>>>>>>>>>> assertion "ribosomes are required for translation to occur",
>>>>>>>>>> and another that supports the assertion "subunit X is
>>>>>>>>>> required for the assembly of a complete ribosome", but no
>>>>>>>>>> single paper that asserts "the ribosomes that actually
>>>>>>>>>> participate in translation have been shown to contain subunit
>>>>>>>>>> X", and so we are left to draw this conclusion by inference,
>>>>>>>>>> reading the two papers in succession. I guess all of this
>>>>>>>>>> violates the principle that there should be a 1:1:1 mapping
>>>>>>>>>> of protein : GO term : publication and evidence code. Here,
>>>>>>>>>> Reactome has silently done the inference so formally you can
>>>>>>>>>> make a 1:1:1 relationship, but there's still a hidden
>>>>>>>>>> multi-step inference in there.
>>>>>>>>>> That said, we live by that hack - our unit of curation is the
>>>>>>>>>> reaction, and I have yet to find the reaction all of whose
>>>>>>>>>> attributes can be annotated without combining evidence
>>>>>>>>>> assertions inferentially from multiple papers.
>>>>>>>>>> Peter
>>>>>>>>>> -----Original Message-----
>>>>>>>>>> From: annotation-bounces at genome.stanford.edu on behalf of
>>>>>>>>>> Midori Harris
>>>>>>>>>> Sent: Wed 6/18/2008 5:02 AM
>>>>>>>>>> To: Karen Christie
>>>>>>>>>> Cc: GO Annotation list
>>>>>>>>>> Subject: Re: [Annotation] annotating ribosomal proteins
>>>>>>>>>> Reactome has identifiers for complexes ... could you use IPI
>>>>>>>>>> with the
>>>>>>>>>> Reactome ID for the ribosome? (E.g. S. cerevisiae ribosome is
>>>>>>>>>> Reactome:141693; H. sapiens is 72500.)
>>>>>>>>>> If this is a stupid idea, I'll bow out quietly ... it's been
>>>>>>>>>> a long, long
>>>>>>>>>> time since I did annotation.
>>>>>>>>>> m
>>>>>>>>>> On Tue, 17 Jun 2008, Karen Christie wrote:
>>>>>>>>>>> As of the 2006 Annotation camp, we agreed that IPI could
>>>>>>>>>>> only be used when
>>>>>>>>>>> you know it is a direct interaction and that you should fill
>>>>>>>>>>> the with column
>>>>>>>>>>> with the directly interacting gene products.
>>>>>>>>>>> Prior to that, I often used to use IPI for the proteins that
>>>>>>>>>>> came down in a
>>>>>>>>>>> complex, and for 'modern' purifications where one protein
>>>>>>>>>>> was tagged, I'd put
>>>>>>>>>>> that one in the with column. But we agreed that it isn't
>>>>>>>>>>> known that this is
>>>>>>>>>>> direct, so we quit doing it. I've been having the same
>>>>>>>>>>> problem with
>>>>>>>>>>> spliceosomal complexes and have become rather unkeen on the
>>>>>>>>>>> decision that IPI
>>>>>>>>>>> could only be used for known direct interactions. It seems
>>>>>>>>>>> that this
>>>>>>>>>>> requirement got added by the people who want to use the IPI
>>>>>>>>>>> with field as a
>>>>>>>>>>> protein-protein interaction database.
>>>>>>>>>>> Anyway, I thought that IPI with the tagged protein seemed
>>>>>>>>>>> like a much better
>>>>>>>>>>> representation of the evidence for the process than IDA.
>>>>>>>>>>> Coming down as part
>>>>>>>>>>> of a complex doesn't seem like a direct assay for process.
>>>>>>>>>>> IPI with one of
>>>>>>>>>>> the proteins in the complex seems like a better
>>>>>>>>>>> representation of what was
>>>>>>>>>>> actually done, but we can't do now that because of the
>>>>>>>>>>> direct interaction
>>>>>>>>>>> requirement.
>>>>>>>>>>> IC also seems a little unsatisfying, since it's not
>>>>>>>>>>> experimental. I don't
>>>>>>>>>>> know, maybe for complexes as well characterized as the
>>>>>>>>>>> ribosome, or the
>>>>>>>>>>> spliceosome, just being in the complex is direct evidence
>>>>>>>>>>> that it's part of
>>>>>>>>>>> the process that the complex is experimentally characterized
>>>>>>>>>>> to be part of...
>>>>>>>>>>> I've been a bit muddled about how best to deal with these
>>>>>>>>>>> ever since the 2006
>>>>>>>>>>> Annotation camp. I liked using IPI better than IDA...
>>>>>>>>>>> -Karen
>>>>>>>>>>> On Tue, 17 Jun 2008, Pascale Gaudet wrote:
>>>>>>>>>>>> Can you then IPI the process?
>>>>>>>>>>>> Doug howe wrote:
>>>>>>>>>>>> I think:
>>>>>>>>>>>> "the IDA is just for the annotation to the complex term
>>>>>>>>>>>> and then use
>>>>>>>>>>>> IC from the
>>>>>>>>>>>> complex term for the Process annotation"
>>>>>>>>>>>> is the way to go.
>>>>>>>>>>>>
>>>>>>>>>>>> Doug
>>>>>>>>>>>>
>>>>>>>>>>>> Karen Christie wrote:
>>>>>>>>>>>> Hi Pascale,
>>>>>>>>>>>>
>>>>>>>>>>>> Rama is looking at the original papers, and
>>>>>>>>>>>> ribosomes and the
>>>>>>>>>>>> processes
>>>>>>>>>>>> of ribosome assembly are probably better
>>>>>>>>>>>> characterized in
>>>>>>>>>>>> cerevisiae
>>>>>>>>>>>> than in any other eukaryote.
>>>>>>>>>>>>
>>>>>>>>>>>> The real issue here is that what has been shown is
>>>>>>>>>>>> that protein
>>>>>>>>>>>> X is
>>>>>>>>>>>> part of a big complex, e.g. the ribosome, for which
>>>>>>>>>>>> the
>>>>>>>>>>>> function is
>>>>>>>>>>>> known. The sum total of the experimental evidence
>>>>>>>>>>>> available for
>>>>>>>>>>>> a
>>>>>>>>>>>> significant number of ribosomal proteins is that
>>>>>>>>>>>> they are
>>>>>>>>>>>> purified as
>>>>>>>>>>>> part of the ribosome complex. So, for component,
>>>>>>>>>>>> it's easy.
>>>>>>>>>>>> This is IDA
>>>>>>>>>>>> evidence that protein X is in the ribosome, or in
>>>>>>>>>>>> the Small
>>>>>>>>>>>> SubUnit
>>>>>>>>>>>> (SSU) or in the Large SubUnit (LSU), or whatever
>>>>>>>>>>>> complex is
>>>>>>>>>>>> characterized.
>>>>>>>>>>>>
>>>>>>>>>>>> But is being in the ribosome considered to be IDA
>>>>>>>>>>>> evidence for
>>>>>>>>>>>> a process
>>>>>>>>>>>> annotation to translation? In one way of looking at
>>>>>>>>>>>> it, the
>>>>>>>>>>>> direct assay
>>>>>>>>>>>> is that it's part of a complex and then you're
>>>>>>>>>>>> assuming that
>>>>>>>>>>>> the
>>>>>>>>>>>> individual protein is involved in translation
>>>>>>>>>>>> because it's in
>>>>>>>>>>>> that
>>>>>>>>>>>> complex. Is this a direct assay for being involved in
>>>>>>>>>>>> translation? Can
>>>>>>>>>>>> we use IDA for a process annotation? or is it a
>>>>>>>>>>>> more accurate
>>>>>>>>>>>> statement
>>>>>>>>>>>> to say that the IDA is just for the annotation to
>>>>>>>>>>>> the complex
>>>>>>>>>>>> term and
>>>>>>>>>>>> then use IC from the complex term for the Process
>>>>>>>>>>>> annotation?
>>>>>>>>>>>>
>>>>>>>>>>>> -Karen
>>>>>>>>>>>>
>>>>>>>>>>>> On Tue, 17 Jun 2008, Pascale Gaudet wrote:
>>>>>>>>>>>>
>>>>>>>>>>>> Hi Rama,
>>>>>>>>>>>>
>>>>>>>>>>>> I think this is a perfect case where one of
>>>>>>>>>>>> us should go
>>>>>>>>>>>> back to the original papers and find
>>>>>>>>>>>> what we all need to ISS to (in which organism
>>>>>>>>>>>> the funtion
>>>>>>>>>>>> and process were shown).
>>>>>>>>>>>>
>>>>>>>>>>>> Pascale
>>>>>>>>>>>>
>>>>>>>>>>>> Rama Balakrishnan wrote:
>>>>>>>>>>>> Hi,
>>>>>>>>>>>>
>>>>>>>>>>>> I have couple of ribosomal proteins to
>>>>>>>>>>>> annotate as
>>>>>>>>>>>> part of the ref-genome curation
>>>>>>>>>>>> project. Turns out that there is no direct
>>>>>>>>>>>> experimental evidence showing that these
>>>>>>>>>>>> proteins are involved in translation.
>>>>>>>>>>>> Almost all
>>>>>>>>>>>> the
>>>>>>>>>>>> studies purify the ribosome
>>>>>>>>>>>> from yeast and identify the subunits by
>>>>>>>>>>>> one or more
>>>>>>>>>>
>>>>>>>>>>>> techniques.
>>>>>>>>>>>>
>>>>>>>>>>>> I can do IDA for CC annotation, that is
>>>>>>>>>>>> straightforward. Is IDA for function
>>>>>>>>>>>> annotation- structural constituent of
>>>>>>>>>>>> ribsomome
>>>>>>>>>>>> okay?
>>>>>>>>>>>> What about BP? I can do IC
>>>>>>>>>>>> from the CC term, but that is not direct
>>>>>>>>>>>> experimental
>>>>>>>>>>>> evidence. What do you all
>>>>>>>>>>>> think?
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks for your time,
>>>>>>>>>>>>
>>>>>>>>>>>> Rama
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> Annotation mailing list
>>>>>>>>>>>> Annotation at geneontology.org
>>>>>>>>>>>>
>>>>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>>>>>>
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> Annotation mailing list
>>>>>>>>>>>> Annotation at geneontology.org
>>>>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>>>>> _______________________________________________
>>>>>>>>>>> Annotation mailing list
>>>>>>>>>>> Annotation at geneontology.org
>>>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Annotation mailing list
>>>>>>>>>> Annotation at geneontology.org
>>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>>>> ------------------------------------------------------------
>>>>>>>>>> This email message, including any attachments, is for the
>>>>>>>>>> sole use of the intended recipient(s) and may contain
>>>>>>>>>> information that is proprietary, confidential, and exempt
>>>>>>>>>> from disclosure under applicable law. Any unauthorized
>>>>>>>>>> review, use, disclosure, or distribution is prohibited. If
>>>>>>>>>> you have received this email in error please notify the
>>>>>>>>>> sender by return email and delete the original message.
>>>>>>>>>> Please note, the recipient should check this email and any
>>>>>>>>>> attachments for the presence of viruses. The organization
>>>>>>>>>> accepts no liability for any damage caused by any virus
>>>>>>>>>> transmitted by this email.
>>>>>>>>>> =================================
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Annotation mailing list
>>>>>>>>>> Annotation at geneontology.org
>>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>>> _______________________________________________
>>>>>>>>> Annotation mailing list
>>>>>>>>> Annotation at geneontology.org
>>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>>>>> _______________________________________________
>>>>>>>> Annotation mailing list
>>>>>>>> Annotation at geneontology.org
>>>>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>>> _______________________________________________
>>>>> Annotation mailing list
>>>>> Annotation at geneontology.org
>>>>> http://fafner.stanford.edu/mailman/listinfo/annotation
>>>
>>>
>
> _______________________________________________
> Refgenome mailing list
> Refgenome at geneontology.org
> http://fafner.stanford.edu/mailman/listinfo/refgenome
More information about the Annotation
mailing list