[Ontology-editors] linking GO terms to docs

[Chris Mungall]

Has anyone done any kind of comparison of annotation tools? The wide  
divergence of tool behavior in terms of visibility of comments could  
be a source of annotator inconsistency. I've also often wondered how  
frequently curators read individual definitions (in actual practice,  
not lip service to practice). It must be very tempting to say "oh, I  
know what that means"...

Anyway, I think we all agree that showing the comments is a Good  
Thing, and tools should aim for this in whatever time frame is  
reasonable.

Back to the original question, I can see two related issues:

- do we need a mechanism for stating comments for an entire sub- 
hierarchy?
- how do we better mutually link documentation on the wiki (and  
tracker?) with the ontology?

For the first issue, I have often thought we should have a concise way  
of showing a definition path and a comment path. Although comments are  
particular to a term, the comments of ancestors may still apply when  
selecting a term for annotation. Thus this would be mostly for use by  
curators rather than end users. Here we are talking about a general  
change in annotation tool behavior rather than adding a new tag to the  
obo file.

For the second, perhaps we could have links going from the ontology to  
web pages (wiki pages or static URLs). But an even better approach  
would be to have something more dynamic, in which comments are  
dynamically fetched from the wiki. This is far more flexible, and  
there is no redundancy and thus less chance of the wiki-ontology links  
becoming stale.

In fact we already have the mechanism for doing this with AmiGO 1.6  
and GONUTS

Have a look on the amigo1.6. release candidate page:
http://goweb-dev.stanford.edu/cgi-bin/amigo/term-details.cgi?term=GO:0007165

Click where it says "community" - you'll see there is an entry that  
links to Jen's signaling pages on our wiki.

Now some details would need to be worked out

- Having two wikis with links between them is not ideal. Not a long  
term problem as we could eventually merge them or use the gonuts  
technology on our wiki.
- Having community comments mixed in with official GO wiki may not be  
to everyone's liking but I think it will be easy to separate these
- This doesn't address the propagation down of comments for  
annotators, but the gonuts API makes it simple for tool developers to  
pull this information dynamically. But that's not so relevant here,  
because annotator-centric comments should continue to go in the  
comments field, and annotators should be hooked in to GO enough to  
know about the signaling thing and other ontology issues

Just to be clear, I'm not talking about replacing the existing  
comments field. This should remain, and fulfil the same purpose it's  
always filled. However, where we have highly fluid information, or  
entire formatted interlinked pages of information, the wiki is the  
appropriate place, and we should have mechanisms to fetch this  
automatically rather than manually maintaining links between the two.

On Mar 25, 2009, at 12:29 PM, Harold Drabkin wrote:

> Yes, I agree; the comments for ALL of them are very useful. So both  
> annotator and user needs to be looking at them as well as the def  
> (in a perfect world 8-|
> h
>
> Alexander Diehl wrote:
>> The point should be made that comments are not just for obsolete  
>> terms, but often help to clarify the meaning of particular terms or  
>> help annotators or users to find related terms which may be more  
>> appropriate. All annotators should pay attention to them,  
>> particularly when using terms that are new to them.
>>
>> See "extracellular space" and "extracellular region":
>>
>> http://www.informatics.jax.org/searches/GO.cgi?id=GO:0005615
>> http://www.informatics.jax.org/searches/GO.cgi?id=GO:0005576
>>
>> -- Alex
>>
>>
>> Harold Drabkin wrote:
>>> On the other hand, anyone can download oboedit and load the GO,  
>>> and set it up to display comments without much difficulty. Once  
>>> set up, it is a excellent browser.
>>> I can't get t Amigo to display an obsolete term, so if someone  
>>> thinks a term was in or should be, they won't get any friendly  
>>> "hints" from Amgio that they might be barking up the wrong tree. 8-(
>>>
>>> h
>>>
>>> Karen Christie wrote:
>>>> While I see your point that it would probably be good to make  
>>>> comments available for both curators and users, changing a tool  
>>>> is a little more involved than "extreme bribing or pestering of  
>>>> the programmer responsible for your browser".
>>>>
>>>> First of all, a database change would be required, to have a  
>>>> place to load this field, before it's even possible to display it  
>>>> via any web interface. While I can put it on a list of  
>>>> suggestions, the decision of whether or when to make such a  
>>>> change would go to SGD's management group who would balance it  
>>>> against other priorities for SGD's database administrator and  
>>>> programmers to make the changes and for curators to test it. If I  
>>>> put it on the list today, I would suspect it could take a year or  
>>>> two before such a change made the priority list to be implemented  
>>>> and actually in production.
>>>>
>>>> While it's great that QuickGO and MGI are ahead of the curve on  
>>>> this particular issue, we should be aware that comments will  
>>>> probably remain really inaccessible to a significant fraction of  
>>>> GO annotators for a while yet.
>>>>
>>>> -Karen
>>>>
>>>>
>>>>
>>>> On Wed, 25 Mar 2009, Emily Dimmer wrote:
>>>>
>>>>> Thanks :-)
>>>>>
>>>>> I'd advise extreme bribing or pestering of the programmer  
>>>>> responsible for your browser - if the comments section is  
>>>>> populated more often, then it becomes important that your tool  
>>>>> is able to properly display this part of an entry.
>>>>>
>>>>> Emily
>>>>>
>>>>> Karen Christie wrote:
>>>>>> It's great that QuickGO already shows comments, but I think you  
>>>>>> may be in the minority there. SGD curators do pay attention to  
>>>>>> definitions, but as we don't even load comments into the  
>>>>>> database, there is no way to see comments in any SGD user or  
>>>>>> curator interface showing GO.
>>>>>>
>>>>>> My point was that I suspect that's not a particularly rare  
>>>>>> situation.
>>>>>>
>>>>>> -Karen
>>>>>>
>>>>>>
>>>>>> On Wed, 25 Mar 2009, Emily Dimmer wrote:
>>>>>>
>>>>>>>
>>>>>>> With regards to the tools - once the information is made  
>>>>>>> available, then it becomes the curators responsibility to  
>>>>>>> properly view the information attached to a term (e.g. some  
>>>>>>> curators may be in the habit of not opening up a GO term to  
>>>>>>> read its definition, but this practice is strongly discouraged  
>>>>>>> by the group). Tool developers should also be encouraged to  
>>>>>>> make the comments section visible for those tools used for  
>>>>>>> curation purposes.
>>>>>>>
>>>>>>> And just to be inflammatory ;-) a curator looking at a term's  
>>>>>>> details in QuickGO already does see the comments section, e.g. http://www.ebi.ac.uk/ego/GTerm?id=GO:0005615
>>>>>>>
>>>>>>> Emily
>>>>>>>
>>>>>>>
>>>>>>> Karen Christie wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I tend to agree with Chris that if we're going to put this  
>>>>>>>> kind of comment/tag in, that it should be added to each term  
>>>>>>>> for which it is relevant, as that seems the only real chance  
>>>>>>>> that it may be seen.
>>>>>>>>
>>>>>>>> Regarding keeping links current, it seems that since they are  
>>>>>>>> to our own wiki, the main thing would be not changing the  
>>>>>>>> wiki url once the comments/tags are put in, and then removing  
>>>>>>>> the comments/tags once that particular area has been finished.
>>>>>>>>
>>>>>>>> While I don't object to them being added, I really wonder how  
>>>>>>>> much they will be seen by the desired target audience. Even  
>>>>>>>> in OBO-Edit, you have to specifically think to look at  
>>>>>>>> comments, and lots of other tools or browsers don't show them  
>>>>>>>> at all. For example, in SGD, I am the only curator/annotator  
>>>>>>>> (out of 10) who uses a method of browsing for terms where it  
>>>>>>>> is even possible to see comments.
>>>>>>>>
>>>>>>>> -Karen
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, 25 Mar 2009, Jennifer Deegan (nee Clark) wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> If we do this I would be in favour of having a curation  
>>>>>>>>> status tag in obo and having defined semantics in terms of  
>>>>>>>>> how it propagates up and down as Chris suggests. This will  
>>>>>>>>> make the information easier to keep up to date. My major  
>>>>>>>>> concern is the amount of work required to keep the links  
>>>>>>>>> current.
>>>>>>>>>
>>>>>>>>> Jen
>>>>>>>>>
>>>>>>>>> Emily Dimmer wrote:
>>>>>>>>>> Hi Chris,
>>>>>>>>>>
>>>>>>>>>> The idea of more fully linking discussions on the GO wiki  
>>>>>>>>>> to the terms was discussed at the last GO Consortium meeting.
>>>>>>>>>>
>>>>>>>>>> Such links provided between terms and GO wiki pages would  
>>>>>>>>>> help curators become more aware of the current ontology/ 
>>>>>>>>>> annotation development efforts. Curators might not consider  
>>>>>>>>>> themselves expert in a topic and so not join a specific  
>>>>>>>>>> working group, but they want to have the option of looking  
>>>>>>>>>> at the current relevant GOC discussions when they are  
>>>>>>>>>> choosing a term for annotation purposes. I strongly feel  
>>>>>>>>>> that there is no point having extensive discussion  
>>>>>>>>>> regarding a set of terms if this conversation hidden in the  
>>>>>>>>>> depths of the GO wiki. If the curator has a concern with an  
>>>>>>>>>> ontology or annotation discussion that appears in the wiki,  
>>>>>>>>>> then text on wiki pages should direct them an appropriate  
>>>>>>>>>> mailing list.
>>>>>>>>>>
>>>>>>>>>> The comment section seems the correct place to put such  
>>>>>>>>>> URLs, as advice on the appropriateness of a term's usage is  
>>>>>>>>>> already included in this section (e.g. extracellular region  
>>>>>>>>>> and extracellular space terms). Therefore GO browsers which  
>>>>>>>>>> support curator annotation activities should already  
>>>>>>>>>> display the comments section for a GO term.
>>>>>>>>>>
>>>>>>>>>> I agree that it would be far better to have such comments  
>>>>>>>>>> propagated in OBO than have the individual tools try to do  
>>>>>>>>>> this, esp. if there are some comments which are appropriate  
>>>>>>>>>> for a whole node of an ontology while others might be  
>>>>>>>>>> specific to a certain term.
>>>>>>>>>>
>>>>>>>>>> Emily
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Chris Mungall wrote:
>>>>>>>>>>>
>>>>>>>>>>> Who is this message aimed at? End-users, annotators, both?  
>>>>>>>>>>> How are people meant to react to this information?
>>>>>>>>>>>
>>>>>>>>>>> I seriously doubt that tools implementors will take the  
>>>>>>>>>>> time write ad-hoc one-off code to propagate this one  
>>>>>>>>>>> particular comment down the hierarchy.
>>>>>>>>>>>
>>>>>>>>>>> If this is important, then why not take the low-tech  
>>>>>>>>>>> approach and (1) email go-friends and (2) propagate the  
>>>>>>>>>>> comment down ourselves and manually remove individual  
>>>>>>>>>>> comments as we fix things.
>>>>>>>>>>>
>>>>>>>>>>> (1) is the most effective at reaching people as not all  
>>>>>>>>>>> browsers necessarily show comments, and when they do it's  
>>>>>>>>>>> only in certain contexts. You could be browsing the tree  
>>>>>>>>>>> or looking at term enrichment results and never see this.
>>>>>>>>>>>
>>>>>>>>>>> If this kind of thing is to become common then we should  
>>>>>>>>>>> figure out a curation status tag in obo, have defined  
>>>>>>>>>>> semantics in terms of how it propagates up and down, etc
>>>>>>>>>>>
>>>>>>>>>>> On Mar 25, 2009, at 7:00 AM, Jennifer Deegan (nee Clark)  
>>>>>>>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Hi,
>>>>>>>>>>>>
>>>>>>>>>>>> The discussion to start overhauling the signaling terms  
>>>>>>>>>>>> is just starting to get going now and we would like to  
>>>>>>>>>>>> make sure all annotators are aware of this. Emily has  
>>>>>>>>>>>> suggested that we might put a pointer to the signaling  
>>>>>>>>>>>> wiki page into the comment field of the signal  
>>>>>>>>>>>> transduction terms to alert users to its existence. She  
>>>>>>>>>>>> says that this comment could then be cascaded down to the  
>>>>>>>>>>>> child terms by the various browser tools.
>>>>>>>>>>>>
>>>>>>>>>>>> If no one has any objections I will add a comment to  
>>>>>>>>>>>> signal transduction on Friday and it will say:
>>>>>>>>>>>>
>>>>>>>>>>>> "The signaling terms in GO are currently being  
>>>>>>>>>>>> overhauled. If you would like to read about the  
>>>>>>>>>>>> discussions, or contribute your ideas or thoughts please  
>>>>>>>>>>>> visit http://wiki.geneontology.org/index.php/Signaling."
>>>>>>>>>>>>
>>>>>>>>>>>> I will also improve that page on the wiki so it is clear  
>>>>>>>>>>>> to people where to go to read about discussions and  
>>>>>>>>>>>> contribute ideas.
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>
>>>>>>>>>>>> Jen
>>>>>>>>>>>>
>>
>
>