[Ontology-editors] GO error: cycle at "integument development"

[Karen Christie]

Hi Donghui,

Thanks for figuring out what happened. I noticed something similar while 
testing the Graph Editor. It will let you make two relationships between a 
term, e.g. one is_a and one part_of, but because the relationships both 
have exactly the same path, the GE effectively only shows one of the two. 
So, if you accidentally created two different relationships between two 
terms, you wouldn't see that in the GE.

-Karen


On Wed, 15 Apr 2009, Donghui Li wrote:

> Hi Karen and all,
>
> Thanks for catching this error.  I'll correct it this afternoon.
>
> I've figured out exactly what happened.
>
> integument development
>   is_a tissue development
>   part_of ovule development
>
> I first created 'integument development' as a child of tissue development by 
> right clicking and choose 'create a new child'.  I then used the graphic 
> editor to add the part_of link between the newly created integument 
> development and the existing term ovule development.  This was done by 
> drawing a line between these two terms, pointing from integument development 
> to ovule development. The default value is always set as 'is_a', I realized 
> this while I was drawing the line/link between these two terms; I then 
> stopped before the line reached the parent term ovule development.  So what 
> happened here was this created a is_a relationship between integument 
> development and itself, although I did not complete the line that links the 
> terms.  I then created the part_of link and that worked.
>
> I normally use graphic viewer to verify term parentage before saving, the 
> graphic viewer does not display this is_a to itself relationship. So the 
> lesson I learned is when using graphic editor to add relationship between two 
> terms, choose the right relationship type before drawing the line/link. I 
> think it's good to add a verification check.
>
> Donghui
>
>
>
>
>
>
> On Apr 15, 2009, at 9:27 AM, Karen Christie wrote:
>
> Hi all,
>
> There is a problem in the current GO file. The term "integument development" 
> (GO:0080060) is listed as an is_a parent of itself. I can fix that later 
> today, but since you added the term originally Donghui, maybe it would be 
> better if you fix it so that you can also check to make sure it has all the 
> parents you intended this term to have.
>
> Also, since it seems like this should never be allowed, for a term to be its 
> own parent, maybe we could also add a verification check to make sure we 
> catch these before committing to CVS.
>
> thanks,
>
> -Karen
>
>
>> On Apr 15, 2009, at 8:33 AM, Shuai Weng wrote:
>> 
>>> There is a bad entry in the current obo file (see below). The GOID (80060)
>>> itself is listed as it's parent - so things go to an infinity loop...
>>> I guess go curators need to fix that. In the mean time, I'll fix the go 
>>> path loading script to check to make sure the parent goid not itself
>>> and try to load the data into SDEV first.
>>> Shuai
>>> ======================
>>> [Term]
>>> id: GO:0080060
>>> name: integument development
>>> namespace: biological_process
>>> def: "The process whose specific outcome is the progression of the 
>>> integument over time, from its formation to the mature structure. 
>>> Integument is one of the layers of tissue that usually covers the ovule, 
>>> enveloping the nucellus and forming the micropyle at the apex." 
>>> [PMID:19054366, PO:0020021]
>>> is_a: GO:0009888 ! tissue development
>>> is_a: GO:0080060 ! integument development
>>> relationship: part_of GO:0048481 ! ovule development
>>> =============
>>> On Wed, 15 Apr 2009, Gail Binkley wrote:
>>>> Thanks Shuai!  No objections from me.
>>>> - Gail
>>>>> I'll look into this and may need to rerun the scripts on fasolt
>>>>> first for testing. Any objections?
>>>>> Shuai
>>>>> On Wed, 15 Apr 2009, Shuai Weng wrote:
>>>>>> I think something is wrong in the obo file.
>>>>>> Shuai
>>>>>> On Wed, 15 Apr 2009, Gail Binkley wrote:
>>>>>>> in both SDEV and SGD with enormous log files.  An example
>>>>>>> DBD::Oracle::st execute failed: ORA-12899: value too large for column
>>>>>>> "BUD"."GO_PATH"."ANCESTOR_PATH" (actual: 241, maximum: 240) (DBD 
>>>>>>> ERROR:
>>>>>>> error possibly near <*> indicator at char 213 in '
>>>>>>>    INSERT INTO BUD.go_path(child_go_no, ancestor_go_no,
>>>>>>>                                generation, ancestor_path,
>>>>>>>                                relationship_type)
>>>>>>>    VALUES(:p1, :p2, :p3, :<*>p4, :p5)
>>>>>>> ') [for Statement "
>>>>>>>    INSERT INTO BUD.go_path(child_go_no, ancestor_go_no,
>>>>>>>                                generation, ancestor_path,
>>>>>>>                                relationship_type)
>>>>>>>    VALUES(?, ?, ?, ?, ?)
>>>>>>> " with ParamValues: :p5='', :p3=35, :p1=28423,
>>>>>>> :p4='80060::80060::80060::80060::80060::80060::80060::80060::80060::80060::
>>>>>>> 80060::80060::80060::80060::80060::80060::80060::80060::80060::80060::80060
>>>>>>> ::80060::80060::80060::80060::80060::80060::80060::80060::9888::48513::4873
>>>>>>> 1::48856::32502::8150', :p2=4284] at
>>>>>>> /usr1/fasolt/www-data/bin/util/cron/../../projects/go/updateGoPath 
>>>>>>> line
>>>>>>> 589.
>>>>>>> What is making the ancestor_path so long?
>>>>>>> - Gail
>> 
>
>
> --
> Donghui Li
> TAIR Curator
>
> http://www.arabidopsis.org
> http://www.geneontology.org
>
> Carnegie Institution
> 260 Panama Street
> Stanford, CA 94305
> U.S.A
>
> Email donghui at stanford.edu
> Tel (650) 325 1521 ext 356
> Fax (650) 325 6857
>
>