Continuing Discussion of Response and Detection Terms

[Midori Harris]

Hello,

What a lot to think about! Thanks to all who have contributed so far. For 
those of us coming into this discussion at this stage, it's very good to 
have all this background, and to see what progress has been made already. 
Now I'd like to try to identify the outstanding problems, in the hope of 
making it easier to grasp what still needs to be decided and done. I'm 
sure I'll miss out a lot of details; I'm not too worried about that just 
now, but do speak up if I miss anything major.

On the topic of which taxa to name in 'response to <organism>' terms, I 
agree that it's an important point, but it's one that can be separated 
fairly cleanly from the other outstanding issues.

Comments to the effect that not all responses are defense responses 
cropped up repeatedly. This may already have been cleared up, but I want 
to make sure there's no lingering doubt: this is NOT one of the problems. 
GO already has 'response to <>' terms for a few taxa, with 'defense 
respons to <>' terms as is_a children, e.g.

  response to bacteria GO:0009617
  -[i] defense response to bacteria GO:0042742

We have no intention of changing this arrangement, whatever is eventually 
decided about 'response to pathogenic <>' terms.

It seems that the biggest outstanding problem is what to do about 
'pathogen' terms. I agree with Alex (and others) that whether one organism 
is pathogenic towards another is highly context- and response-dependent, 
and find myself somewhat swayed by the arguments against including 
'pathogen' in term names. I'm not going to argue against 'potential 
pathogen' terms, however, if they emerge as the best compromise.

For this example:

  There are many plant responses which are triggered by pathogen in
  order to make the plant more susceptible.

GO has devised a way to accommodate this sort of situation at the level of 
annotation. The details are kind of outside the scope of this discussion 
(and it's clunky, and I've just noticed that we ought to improve the 
documentation on it), but the fact that we have a way to represent one 
organism co-opting a gene product made by another in annotations is 
relevant to the question of what has to be represented in the ontology. 
(I'll fill ni details later upon request.)

Given that we can separate out cases where a pathogen uses a host gene 
product, we seem to be left with at least substantial overlap between 
(e.g.) 'response to <> pathogen' and 'defense response to <>'. Even if 
they are not exactly synonymous, they may be similar enough for us to use 
non-exact 'synonyms' instead of separate GO terms for the two, e.g. (in 
partial OBO format):

  id: GO:0042742
  name: defense response to bacteria 
  related_synonym: response to bacterial pathogen
  is_a: GO:0009617 ! response to bacteria 

One tangential note: there has been nothing at all on the main GO mailing 
list about tentative dates for a content meeting.

Midori