Ontology meeting 2014-03-13

From GO Wiki
Revision as of 12:15, 13 March 2014 by TanyaB (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Attendees: Paola, DavidOS, DavidH, Chris, Heiko, Tanya, Harold

Minutes: Tanya

New release of OBO-Edit

Release 2.3.1-b1 is available for testing (https://sourceforge.net/projects/geneontology/)

The only change is support for annotation properties (e.g. neverintaxon).

Everyone will need to switch to 2.3.1 before moving taxon constraints (TCs) into editors' file: https://www.ebi.ac.uk/panda/jira/browse/GO-267#comment-92325

We need to do some more testing of 2.3.1b first, but we don't anticipate any issues.

Where are we with this?

 Editors to download and test. Pls. do ASAP. Release will be done right after meeting.

'colocalizes with' nuclear chromatin

 Experiment: gene product is bound to nuclear chromatin, specifically the pol3 promoter of gene X
 Proposed changes: need new GO cell component term , chromosomal region/segment (domain), pol3 promoter (SO term), new relationship 'coincident with', of 
 Annotation: Genus/col5: GO:new ! chromosome segment, col16: col16: coincident_with(SO:nnnn)! pol3 prom, part_of(GO:0000790)! nuclear chromatin,     coincident_with gene_product_id (MGI:MGI:nnnnnn)
 Ontology edits: need to change complex terms that are is_a children of nuclear chromatin to be part_of - open JIRA issue
 New relationship type: add 'coincident with' to RO with generic definition - DavidOS to add to RO
 Related problem: cytoskeleton child terms that are is_a and should probably be part_of - open JIRA issue


Carried over from annotation call on Tuesday. Create new component terms. Discuss.

http://gocwiki.geneontology.org/index.php/Annotation_Conf._Call,_March_11,_2014#Curation_question

My point was that adding a new relation doesn't help as you're still making a statement

SUBUNITX colocalizeswith some (nuclearchromatin AND some pol3promoter)

The subunit needs to be connected directly with the promoter to ensure the interpretation is as strong as intended.
 I recommended adding new terms to the ontology that are parts of chromatin that are cognates of the SO sequences.
 I also suggested that =has_part would be logically valid but weaker than the intent.

However, I then realized I wasn't completely sure on the strict meaning of nuclear chromatin in GO 
- does it mean the entirety or a portion? This is one of those situations where it's actually important 
to be clear.

I found the highlighted placements a little odd:

 is_a GO:0044428 ! nuclear part
  is_a GO:0044454 ! nuclear chromosome part
   is_a GO:0000790 ! nuclear chromatin
    is_a GO:0005719 ! nuclear euchromatin
    is_a GO:0005720 ! nuclear heterochromatin
     is_a GO:0001739 ! sex chromatin
     is_a GO:0005724 ! nuclear telomeric heterochromatin
     is_a GO:0005725 ! intercalary heterochromatin
     is_a GO:0031618 ! nuclear centromeric heterochromatin
     is_a GO:0098578 ! condensed chromatin of inactivated sex chromosome
     is_a GO:0098580 ! chromatin of active sex chromosome
     is_a GO:1902377 ! nuclear rDNA heterochromatin
    is_a GO:0070209 ! ASTRA complex*
    is_a GO:0070211 ! Snt2C complex*
    is_a GO:0070823 ! HDA1 complex*
    is_a GO:0098580 ! chromatin of active sex chromosome
Maybe these are mistakes, and the intention is part_of?

But it seems the intention is that NC can be any portion - which allows us to make a class 
expression with the intended meaning

(nuclearchromatin and hasdnapartconsistingonlyof some RNAPol3_promoter)

Not suggesting this actual relation, just to demonstrate the possibility.

I think it's cleaner to flip around the class expression and make the promoter region the genus. 
Or alternatively, to make the genus (col5 in a GAF) some generic small chromosome region and 
include two part_of differentia, to NC and to the promoter.

There's a wider issue here about the gap between the interpretation a reasoner might make and
 those a curator might make, especially if the curation tool doesn't make use of reasoning to test 
the generated expression. I can see how this might be frustrating for CC annotations where the 
intent is to make an expression that is part_of the GO CC class.

Anyway, I said we would discuss on thursday.

see also - discussion in thread "Re: [go-discuss] annotation extensions: relation to use for localization to chromosome at specific sequence feature" from Jan/Feb 14


AOB

Brief discussion on presentations for GOC Saturday meeting and slides.