'with' field contents for IPI annotations

[Evelyn Camon]

Hi David,

The main sequence does not always represent isoform 1 but it usually 
does. TrEMBL records create problems for us also and so we annotate 
isoforms only when they are Swiss-Prot records. I am ccing 
Michele/Eleanor to confirm. At the time you look at a TrEMBL record the 
sequence might represent a specific isoform...later it will be merged 
with other entries and a different perhaps longer sequence might be 
chosen as the representative sequence in the Swiss-Prot record. The 
TrEMBL identifier might then become secondary to this new accession I am 
not aware of a way of linking the new isoform ids to the old TrEMBL 
accessions but it may be possible???? Michele or Eleanor can confirm. I 
believe once an isoform ID is assigned it is stable??(Ele).

Some minutes of our discussion on the topic for your information.

Evelyn

****************
The information that is manually annotated to UniProtKB/Swiss-Prot 
record represents the characterisation of all splicing variants. However 
each splicing variant is designated a unique isoform id e.g 
(P12345.1)and it is possible using a script to separate the sequences 
into those that represent the individual isoforms.

Up to Novemeber 2005 GOA had no method of annotating directly to splice 
variants and so all possible annotations were assigned to the main 
protein accession i.e P12345. This meant that it was impossible for 
biologists to extract the specific variant GO annotation. The ability to 
annotate to splice variants allows GOA also to experimentally validate 
the splicing variants as in UniProtKB they are not all experimentally 
verified.

DECISION at GOA Meeting 7-11-2005

If you know specific variant GO annotation then annotate to the 
variant/isoform accession(P12345.1).

If you don't know which specific variant GO annotation should apply then 
annotate to the main entry(P12345). These are ok to generally transfer 
to other species by ISS.

If you know specific isoform has a match in another species then can 
transfer by ISS from isoform to isoform (wait until David updates tool)
Notes: No need to try an summarise GO annotation to higher level terms 
from specific variants in main entry. Eleanor made point that isoform.1 
is not always the sequence on display in UniProtKb.
****************




David Hill wrote:
> Actually, if an author is looking at a specific isoform, it is usually 
> easy to tell which Uniprot-# it goes with by looking at the sequence. 
> The probelm I run across is when the isoform corresponds to a Trembl 
> record. Is that record just rolled into the generic uniprot id or does 
> it actually become associated with the correct isoform?
> 
> David
> 
>>
>>
>> We will also accept UniProt Isoform Ids (P00001.1, P00001.2) if you 
>> can figure out which is the correct one in the paper and its has an id 
>> in UniProt..bit of an awkward one at the moment.
> 
> 
> 
> 
> 


-- 
Evelyn Camon
GOA Coordinator
Senior Scientific Curator
European Bioinformatics Institute
Tel:01223-494465
Fax:01223-494468
E-mail: camon at ebi.ac.uk
URL: http://www.ebi.ac.uk/goa