[go] AmiGO beta announcement

[Chris Mungall]

Ben is right, we don't have public-friendly versioning procedures.

There is a proposal to improve this, see:
http://gocwiki.geneontology.org/index.php/Versioning_Proposal

This will take some work.

In the meantime, some guidelines:

The CVS revision number is *not* a good guide as this is file/cvs- 
dependent, and GO is available from at least 3 cvs locations: go and  
go_edit on the GO CVS, and also on the obo.sf.net cvs.

However, the cvs revision is useful for fine-grained versioning.  
However, this only applies to the editors version. When we move to a  
*true* editors version of the ontology, early next year, we can set  
up a procedure for ensuring this is the case - we will have a script  
that copies the editors version to the public version, and modifies  
the line that says:

remark: cvs version: $Revision: 5.622 $

to say something like:

remark: editors cvs version: Revision: 5.622

which will *not* be auto-amended by cvs

this will also go into the database as metadata for a release

I will post a more detailed plan early next year

For non-fine-grained versioning, the database counts as a release.  
The release ID is of the form YYYY-MM-DD, and corresponds exactly to  
what was in cvs on that date, and can be extracted using the  
procedure below.

On Dec 20, 2007, at 9:23 AM, Benjamin Hitz wrote:

>
> There's an ontology version?  You mean the CVS revision number?
> It's updated "hourly" ... and the goterm database is loaded daily.
>
> The way the loading works is that it checks out the ontology (and  
> gas, etc.) with a "sticky date".
>
> For the (current, not yet finished) golite
> mysql at goad ontology]$ cvs status -v gene_ontology.edit
> ===================================================================
> File: gene_ontology.obo Status: Up-to-date
>
>    Working revision:    4.377
>    Repository revision: 4.377   /anoncvs/go/ontology/ 
> gene_ontology.obo,v
>    Sticky Tag:          (none)
>    Sticky Date:         2007.12.20.08.00.00
>    Sticky Options:      (none)
>
>    Existing Tags:
>         metabolism_branch_oct_04        (branch: 3.557.2)
>
>
> So anyone can get the exact file used with >cvs update -dA -D  
> 2007-12-20
>
> We use the date stamp so that it's easier to recreate a load at a  
> different time.
>
> Ben
> On Dec 20, 2007, at 6:19 AM, Amelia Ireland wrote:
>
>> Back in Gotham City, Valerie Wood wrote:
>>
>>> I agree with David, most analysis which use GO are not reproducible
>>> because details about GO data (annotations and ontology) and  
>>> software
>>> used are ommitted. I think we should provide this info at the top  
>>> of the
>>> results output. Perhaps the load date for now, until we have proper
>>> versioning?
>>
>> How about something saying 'Cite this data', with information  
>> about the
>> ontology version and the date when the annotations were updated?  
>> We should
>> also update the GO website page about citing GO to state that you  
>> should
>> specify this data when you use/publish data from AmiGO.
>>
>> Chris / Ben - can we get the ontology version data put into the  
>> database
>> as well as the date when the annotations were updated?
>>
>> -- 
>> Amelia Ireland
>> GO Editorial Office,
>> European Bioinformatics Institute, UK.
>> Carbon neutral driving: http://www.targetneutral.com/TONIC/index.jsp
>>
>
> --
> Ben Hitz
> Senior Scientific Programmer ** Saccharomyces Genome Database ** GO  
> Consortium
> Stanford University ** hitz at genome.stanford.edu
>
>
>
>