[go-helpdesk] membrane v plasma membrane

[Susan Tweedie]

Dear Christian

On 21 Jun 2012, at 21/Jun/2012 15:37:39, Christian Moritz wrote:

> Dear Susan Tweedie,
>
> thank you for your very fast answer.
> I used AmiGO (especially the Gene Product Annotation Overview) to  
> search for the GO terms.
>
> And I can give you an example of my problem. (I hope the links work.)
>
> The first link shows a protein, for which both "membrane" and  
> "plasma membrane" was annotated.
>
> http://amigo.geneontology.org/cgi-bin/amigo/gp-assoc.cgi?gp=RGD:1306991&session_id=9487amigo1340183832
>
>
> This second link shows a protein, for which only "plasma membrane"  
> was annotated.
>
> http://amigo.geneontology.org/cgi-bin/amigo/gp-assoc.cgi?ont=cellular_component&evcode=all&gp=RGD%3A735038&session_id=8014amigo1340288081
>
>
> I'm interested in the difference of these two strategies: showing  
> only the direct annotation in case #1 or showing a parent term  
> additionally (case #2) although using the same search strategy.

I  now understand that this difference reflects the underlying data  
rather than your search strategy. When you view the data in this way  
all of the annotations are direct. The confusing feature is that in  
the first case membrane and plasma membrane have both been annotated  
directly based on the same evidence (i.e. from sequence orthology with  
the mouse protein Lamp5 Q9D387).

The rat database (RGD) made these annotations based on an automated  
pipeline described here: http://rgd.mcw.edu/rgdweb/report/reference/main.html?id=1624291 
. Their annotations reflect the fact that the mouse protein has  
distinct manual annotations to both plasma membrane and membrane based  
on experimental evidence. It is not unusual for this to happen -  
different papers provide evidence at different resolutions and  
existing annotations are not necessarily removed when more detailed  
information becomes available.

The unusual thing about this example is that the source mouse  
annotations were based on the same publication (PMID:17215451). It is  
not typical for a curator to annotate to both a specific and general  
term from the same paper as the information is redundant.  I can only  
assume that the paper showed that the protein was in both the plasma  
membrane and another unnamed membrane so they tried to capture this  
second location with the generic term.

In the second example, the rat annotation to plasma membrane is  
inferred from sequence orthology with human CCT3 - there is no direct  
annotation to membrane for that protein.

I hope this answers your question.

Best wishes,
Susan

>
>
> Kínd regards,
> Christian
>
>
>
>
>
>
>
> On Thu, 21 Jun 2012 11:51:30 +0100
> Susan Tweedie <sart2 at gen.cam.ac.uk> wrote:
>> Dear Christian
>> Thank you for contacting the GO Consortium with your question.
>> Your understanding is correct. Plasma membrane is a more specific  
>> type of membrane (an is_a child term of membrane) so all proteins  
>> annotated with plasma membrane are inferred from the hierarchy to  
>> be associated with membrane.
>> I am not sure what tool you used to do your analysis so it is  
>> difficult to trouble-shoot the problem without more information.  
>> However this situation could arise if the method you are using is  
>> only considering direct annotations with these terms rather than  
>> making use of the hierarchy to infer annotations to the parental  
>> terms. The underlying database of associations between terms and  
>> proteins contains only those that have been made directly to a  
>> specific term - the inferences from the hierarchy are not  
>> explicitly listed.
>> So if you search with one of your proteins associated with plasma  
>> membrane and view the associations you would expect to find at  
>> least one entry with plasma membrane but there may not be any  
>> direct annotations with membrane. However if you use tools such as  
>> our Term Enrichment or GO slimmer to analyze your proteins then the  
>> hierarchy will be taken into account.
>> If that doesn't answer your question please get back to us.
>> Best wishes,
>> Susan Tweedie
>> On 21 Jun 2012, at 21/Jun/2012 10:27:09, go-helpdesk-bounces at lists.stanford.edu 
>>  wrote:
>>>
>>> From: christian.moritz at biologie.uni-kl.de
>>> Date: 21 June 2012 10:07:06 BST
>>> To: go-helpdesk at mailman.stanford.edu
>>> Subject: GO Help query (from website)
>>>
>>>
>>>
>>> Email: christian.moritz at biologie.uni-kl.de
>>> Name: Annotation (from Christian Moritz)
>>> Text: Dear GO team,
>>> I have a (perhaps rather basic) question on the terms.
>>> In my analysis of a protein set, 125 out of about 600 ptoteins  
>>> belong to the term "plasma membrane". But only 59 out of the 600  
>>> belong to the term "membrane".
>>> Now I'm wondering how the hierarchy works: shouldn't all plasma  
>>> membrane proteine be included in membrane? Or do I have to add the  
>>> 2 numbers to get the complete number of membrane proteins?
>>>
>>> To understand this would help me alot.
>>> Thanks and best regards,
>>> Christian
>>>
>>>
>>>
>> Susan Tweedie
>> FlyBase GO curator
>> s.tweedie at gen.cam.ac.uk
>
> --
>
> Christian Moritz
> Diplombiologe
>
> Doktorand
> Neuroproteomik-Labor
> Abteilung Tierphysiologie/AG Friauf
> Fachbereich Biologie
> TU Kaiserslautern
>
> http://www.uni-kl.de/wcms/tierphys.html
> Tel.: 0631 205 5004
>
> Erwin-Schroedinger-Straße 13
> D-67663 Kaiserslautern
> _ _ _ _ _ _
>
> Ph.D. student
> Neuroproteomics lab
> Division of animal physiology/Friauf group
> Department of biology
> TU Kaiserslautern
>
> http://www.uni-kl.de/wcms/1768.html?L=1
> Tel.: (0049) (0)631 205 5004
>
> Erwin-Schroedinger-Str. #13
> D-67663 Kaiserslautern
> Germany
>
> Sparen Sie pro Seite ca. 200 ml Wasser, 2 g CO2 und 2 g Holz:
> Drucken Sie daher bitte nur, wenn es wirklich notwendig ist.