Search Mailing List Archives

Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[Obo-phenotype] GO vs Phenotype curation

Chris Mungall cjm at
Mon Jan 22 12:11:35 PST 2007

I think we will have various kinds of overlap and redundancy. I think  
it will be valuable to capture it all (though tools like phenote will  
of course be able to help curators and suggest annotations based on  
rules and statistical associations). This isn't any different from GO  
(eg transcription => nucleus), although due to the nature of  
phenotype annotation we will have a wider variety of granularities  
and perspectives.

For any genotype <gt> that has an allele for <g> where <g> encodes <gp>:

IF E= <p> AND <p> is_a biological_process
AND the phenotype is abnormal
THEN the wild type gene product participates_in <p>

(note: this rule may be correct, but it may not be complete; it may  
be possible to use the Q to make a more specific GO annotation)

If E= <a> AND <a> is_a anatomical_entity (including whole cells)
AND the phenotype is abnormal
THEN then wild type gene product participates in <p>
WHERE <p> = development_of <a>

The cyclops example is interesting. If a cyclops has a single  
perfectly functioning good eye would we really say:

  E= GO:vertebrate_eye_development[*] Tag= abnormal

Surely the eye development executed as normally as can be expected,  
it just had the bad luck to be executed as a part_of a larger  
assemblage of processes such as cranio-facial development? (no  
suitable term in GO). Perhaps we could also say the overall facial  
development program execution was abnormal in that it had only one  
eye development process instance as sub-part, rather than two. I'm  
guessing curators would find that a strange way to annotate.

Of course, the real subpart of the facial development process is  
likely to be some upstream process involved in laying down some kind  
of left-right patterning in a precursor (I'm guessing - don't know  
much about Cylcopeanism). But this inference may not be supported by  
the data in the paper. In fact there may not be any developmental  
data at all; in which case, why not annotate:

   E= FMA:Face
   E2= FMA:Eye

(we need to fix this part of PATO -see also http:// for why it's better to  
state it this way than having E= Eye - this is similar to the  
spermatocytes devoid of asters example)

Here you're stating exactly what you are seeing.

Then at a later time someone may do a different experiment with the  
same gene (or the orthologous gene in another species - or an  
upstream gene.....) and observe at a more detailed level a disruption  
in development during an earlier stage of facial development -  
resulting in both a phenotype annotation and a GO annotation of the  
wildtype. Which doesn't make the first phenotype annotation redundant  
- it's all fantastic data for seeding the kinds of analysis that are  
very run of the mill for GO right now but will be a fair bit more  
complicated for phenotype annotation.

So basically - err on the side of redundancy and when we have a  
decent amount of data lets try some rule mining to see what the best  
ways of transferring annotations across are..


On Jan 22, 2007, at 9:26 AM, Pankaj Jaiswal wrote:

> Sorry for jumping in. The GO and phenotype annotations are  
> essentially the same thing. Only difference is that the way  
> annotations are associated/assembled in the current GO association  
> tables, except for the quality aspect that is new and comes from  
> PATO. In the following example entity is the 'GO_process: eye  
> development'.
> e.g.
> object: allele-A | GO_process: eye development | Quality: abnormal | 
> Code: IMP | Evidence: PMID-xxxxx
> The same annotation gets percolated to the gene as well because of  
> its lineage.
> -Pankaj
> Doug howe wrote:
>> Michael, Consider an alternate example that uses a GO term as the  
>> entity in the phenotype annotation: A mutation in gene A results  
>> in cyclopia. One could annotate gene A with the GO term 'eye  
>> development' by IMP and the allele of gene A with a phenotype  
>> annotation like entity: eye development + quality:abnormal No? - 
>> Doug Michael Ashburner (Genetics) wrote:
>>> Doug I am not sure I see this as a major problem. Let us imagine  
>>> you have a mutation in gene A which lacks the dorsal fin. You  
>>> _might_ deduce from this (or an author might) that it is  
>>> appropriate to annotate the corresponding gene as having the GO  
>>> process fin morphogenesis with an IMP evidence code. That  
>>> annotation is to the protein (the gene being a proxy for this).  
>>> But then the particular _allele_ studied would have a PATO  
>>> annotation entity: fin; quality: absent [or whatever, I am just  
>>> giving the general idea] These annotations do not seem to be  
>>> redundant to me Michael
>> --------------------------------------------------------------------- 
>> ---- Take Surveys. Earn Cash. Influence the Future of IT Join  
>>'s Techsay panel and you'll get the chance to share  
>> your opinions on IT & business topics through brief surveys - and  
>> earn cash 
>> page=join.php&p=sourceforge&CID=DEVDEV  
>> _______________________________________________ Obo-phenotype  
>> mailing list Obo-phenotype at https:// 
> -- Pankaj Jaiswal G-15, Bradfield Hall Dept. of Plant Breeding and  
> Genetics Cornell University Ithaca, NY-14853, USA Ph.  
> +1-607-255-3103 / 4199 fax: +1-607-255-6683
> ---------------------------------------------------------------------- 
> ---
> Take Surveys. Earn Cash. Influence the Future of IT
> Join's Techsay panel and you'll get the chance to  
> share your
> opinions on IT & business topics through brief surveys - and earn cash
> page=join.php&p=sourceforge&CID=DEVDEV________________________________ 
> _______________
> Obo-phenotype mailing list
> Obo-phenotype at

More information about the go-discuss mailing list