WPWG-20080122-transcript

From GO Wiki
Jump to: navigation, search

Conversation with #AmiGO

(08:33:25 AM) Logging started. Future messages in this conversation will be logged.
(08:33:33 AM) sjcarbon: sorry--disk was defragging
(08:34:59 AM) gwg: Hi Seth, hi Eurie

(08:35:47 AM) eurie: morning
(08:35:54 AM) eurie: pretty sparse today, huh?
(08:36:23 AM) sjcarbon: last weeks AmiGO meeting was completely empty. I wonder if there is a problem with the email...
(08:36:32 AM) gwg: yeah... have people forgotten it's on?

(08:36:46 AM) rama [rama@c-24-7-123-136.hsd1.ca.comcast.net] entered the room.
(08:36:57 AM) rama: Hi, sorry I am late
(08:37:13 AM) gwg: I got the email about the WPWG meeting
(08:37:19 AM) gwg: Hi Rama

(08:37:38 AM) sjcarbon: gwg: hmmm...
(08:38:02 AM) rama: Hi
(08:38:05 AM) sjcarbon: Hey!
(08:38:19 AM) gwg: there isn't actually very much to discuss from the agenda

(08:38:20 AM) sjcarbon: I'm sending a reminder email out now.
(08:38:43 AM) rama: I have a question
(08:39:04 AM) gwg: when is AmiGO going to be released?
(08:39:04 AM) rama: about the display of qualifiers for annotations..

(08:39:13 AM) gwg: oh yes
(08:40:09 AM) rama: was there a consensus on where the qualifier shd be placed?
(08:41:16 AM) gwg: no, not really, although the display with the qualifier shifted over [next to the evcode column] seemed less controversial than the original placement
(08:42:09 AM) sjcarbon: it's probably better 'cause it helped find that error in the database.

(08:42:39 AM) eurie: but the qualifiers are used to qualify the GO term and not the ev code
(08:42:50 AM) rama: depending on the release date, I think we should offer few other options and get comments...I am not particularly happy about placing it before the ev.codes.
(08:42:50 AM) gwg: I know
(08:43:08 AM) gwg: it's a bit of a lesser of two evils really

(08:43:32 AM) eurie: the bigger issue from reading that thread is the display of number next to the GO term when there are qualifiers present
(08:43:57 AM) eurie: when it says "54 gene products" (I forget the exact wording right now)
(08:43:59 AM) rama: yes i agree with eurie

(08:44:06 AM) gwg: that was Alex's problem with the original placement
(08:44:21 AM) eurie: next to a GO term with a qualifier, it's not accurate
(08:44:55 AM) eurie: i think going back to the more generic "view associations" that is currently in production would address that ambiguity

(08:45:15 AM) gwg: but then people requested the number of associations to be shown
(08:45:35 AM) gwg: (sorry, please translate that sentence into English at your leisure)
(08:46:03 AM) pascale [pgaudet@modemcable236.104-200-24.mc.videotron.ca] entered the room.
(08:46:13 AM) gwg: bonjour!

(08:46:16 AM) pascale: hi!
(08:46:28 AM) eurie: yeah, we had a huge discussion at SGD about that exact issue when we were displaying our own qualified annotations
(08:46:47 AM) eurie: we currently have separate counts for the unqualified GO term and the qualified GO term - and they are split into different tables
(08:46:58 AM) gwg: obviously the best situation would be if we had masses of screen real estate and the link to the associations was somewhere well away from any 'NOT' that might be present

(08:47:44 AM) gwg: can you give us a link...?
(08:47:47 AM) rama: <a href="http://db.yeastgenome.org/cgi-bin/GO/goTerm.pl?goid=5816">http://db.yeastgenome.org/cgi-bin/GO/goTerm.pl?goid=5816</a>
(08:50:54 AM) rama: if you scroll down a little bit on the URL i provided, you will see that we have two tables, one for the term and one for the term + qualifier
(08:51:02 AM) rama: there are anchors to this table

(08:51:16 AM) gwg: is the colocalizes-with data displayed on both pages for a reason?
(08:51:44 AM) rama: hmmm. both pages? which two pages are you referring to?
(08:52:20 AM) gwg: results 1 - 30 and then results 31 - 42
(08:53:07 AM) rama: no....the pagination is for the annotations to the Term only

(08:53:14 AM) gwg: oh right, ok
(08:53:38 AM) rama: I see your confusion though
(08:55:21 AM) rama: Do we have other ideas for displaying qualifiers?/
(08:55:37 AM) gwg: hmmm, it is certainly a lot clearer displayed in that manner

(08:55:43 AM) pascale: I think it's nice to split it in two tables
(08:56:26 AM) gwg: we could potentially have six tables per term tho
(08:56:35 AM) gwg: with the combinations of qualifiers
(08:56:48 AM) eurie: y

(08:57:06 AM) eurie: oops.. do any groups currently have 2 qualifiers?
(08:57:06 AM) gwg: or do qualifiers not get combined?
(08:57:31 AM) gwg: I think there's a couple of HGNC annots to 'not contributes to'
(08:57:42 AM) rama: wow...that is a loaded annotation

(08:57:45 AM) gwg: or not colocalizes with
(08:57:54 AM) eurie: spindle pole body (GO:0005816) is a good example because there are annotations directly to the term as well as using "colocalizes with" and "NOT" (separately)
(08:57:57 AM) gwg: well, you know these human annotators! ;)

(08:58:21 AM) eurie: [interesting that the colon zero gets translated to a surprise face....]
(08:59:02 AM) gwg: where are the not annotations? are they displayed somewhere else?
(08:59:28 AM) rama: she meant on the AmiGO page (not SGD)
(08:59:29 AM) rama: <a href="http://amigo.geneontology.org/cgi-bin/amigo/go.cgi?search_constraint=terms&query=5816&view=assoc">http://amigo.geneontology.org/cgi-bin/amigo/go.cgi?search_constraint=terms&query=5816&view=assoc</a>

(08:59:48 AM) rama: eurie, is that right?
(08:59:53 AM) eurie: oh, yeah, thanks for translating rama
(09:00:13 AM) eurie: i was looking at the amigo page for a good example of a page where there could be multiple different tables
(09:02:16 AM) rama: shd we go back to the drawing board and think about this?

(09:02:57 AM) gwg: we should probably get more opinions as we have a slight SGD bias today! ;)
(09:03:10 AM) rama: sure
(09:03:35 AM) gwg: I think one thing we might want to think about is hiding the NOT annotations from naive users
(09:04:01 AM) sjcarbon: also, should we keep it as now, or revert to the old view while this is being hashed out?

(09:04:01 AM) rama: or may be provide it one click away
(09:04:42 AM) rama: Seth, what is the Toy version pointing to? and what is the dev version pointing to..
(09:05:11 AM) rama: it will be good to see the two options to come with newer displays
(09:05:58 AM) gwg: IIRC toy has the qualifiers next to the ev codes, and dev has them elsewhere

(09:06:13 AM) eurie: or maybe have the gene list annotated using NOT be arrow toggle-able [looking at the arrow toggle-able Filter display]
(09:06:20 AM) gwg: we could put the qualifiers in the last column, but I think they would be more likely to be missed there
(09:06:26 AM) eurie: oh, actually, toy has the qualifiers before the term
(09:06:30 AM) sjcarbon: <blink>

(09:06:34 AM) eurie: <a href="http://toy.lbl.gov:9006/cgi-bin/amigo/gp-assoc.cgi?gp=SGD:S000006095">http://toy.lbl.gov:9006/cgi-bin/amigo/gp-assoc.cgi?gp=SGD:S000006095</a>
(09:06:54 AM) eurie: uh... hmm.. colon S 0
(09:06:58 AM) gwg: you need toy:lbl.gov:9012 for the latest version
(09:06:59 AM) sjcarbon: :9012 is the current CVS

(09:07:44 AM) rama: okay...we have 9012 and 9006 to compare
(09:08:57 AM) rama: Action item?
(09:09:43 AM) sjcarbon: compare the two and come up with better ways of displaying the qualifiers?
(09:09:57 AM) gwg: essentially what you're doing is choosing how to sort your associations - maybe if we enabled sorting associations by qualifier, so then people could choose to have all the qualified annotations lumped together?

(09:10:17 AM) ben [hitz@bourbon.Stanford.EDU] entered the room.
(09:10:18 AM) gwg: [please translate that sentence into proper English at your leisure]
(09:10:30 AM) ben: sorry, I thought we had a phone call, and was on the train
(09:10:40 AM) sjcarbon: no problem.

(09:10:54 AM) rama: it is also the counts issue GWG
(09:11:13 AM) rama: Hi ben
(09:11:43 AM) gwg: yeah... I can just see people getting all confused about why a term has a certain annotation count in one part of the page, and a different one in another part of the page...
(09:12:25 AM) sjcarbon: and it might mix badly with other kinds of sorting and grouping.

(09:13:41 AM) rama: here are the issues that we need to keep in mind: Counts (direct annotations and with qualifier), combination of qualifiers
(09:14:16 AM) gwg: another thing is that most of the gp counts aren't calculated on the fly, they're put into the database at load time, and they don't take into account the qualifiers contributes-to and colocs-with
(09:14:35 AM) rama: I was just about to ask that.
(09:15:25 AM) gwg: we'd need to ask Chris what he thinks about splitting up the gp counts by qualifier

(09:15:31 AM) rama: hmmm..that makes it hard..
(09:15:49 AM) sjcarbon: action item for me.
(09:16:12 AM) gwg: also, is it useful to know the counts for 'contributes-to' and 'colocalises-with'?
(09:16:20 AM) rama: and NOT

(09:16:32 AM) eurie: i'm not sure -
(09:16:34 AM) ben: I think it would require a minor database change
(09:16:35 AM) gwg: with 'not' annotations, it's obvious that it shouldn't be included in the counts for a term
(09:16:35 AM) pascale: I think so

(09:17:00 AM) eurie: i think going back to the "view associations" version that's in production would be fine
(09:17:03 AM) ben: or wait... it would be easy as long as the gene_product_count table just referred to "unqualified" counts
(09:17:31 AM) ben: but other wise, you need to cross index qualifers (which are themselves terms, not constants) and ...

(09:17:42 AM) eurie: at least until all these other issuse have been worked out.
(09:17:43 AM) rama: I am starting to lean towards the production version too...(with Eurie)
(09:17:55 AM) rama: since we need to think about DB changes..
(09:18:05 AM) eurie: Did the person who suggested the counts for the all the qualified annotations give a reason why?

(09:18:14 AM) gwg: I don't even remember who it was!
(09:18:22 AM) pascale: how are the counts generated then? can you not query based on whether/which qualifier is present
(09:18:23 AM) gwg: it might be in a SF request somewhere
(09:18:58 AM) gwg: the counts are done when the db is being generated

(09:19:17 AM) ben: query would be very slow, that is why they are stored in a separate table (gene_product_count) I guess
(09:19:22 AM) pascale: ok
(09:19:23 AM) gwg: they currently exclude NOT annots, but include other qualifiers
(09:20:20 AM) gwg: ?we could easily change the code so that it excluded other qualifiers too

(09:20:54 AM) gwg: but then what if there are other people who consider that a 'contribs-to' association is equivalent to a direct association?
(09:21:36 AM) pascale: can they not add the counts? it seems easier than the other way around
(09:22:33 AM) gwg: hmmm
(09:23:17 AM) rama: may I suggest that we start a specs page for this issue on the wiki, present some options at the next GOC meeting?

(09:23:25 AM) gwg: yes, good idea
(09:23:30 AM) rama: and go with the current production version for this relase?
(09:23:39 AM) rama: I meant release?
(09:23:59 AM) gwg: I think we should go with the current toy version for this release

(09:24:33 AM) rama: toy:9006 version?
(09:24:39 AM) gwg: mainly because I have a memory of someone pestering for those GP counts
(09:25:30 AM) gwg: I think toy:9012 is marginally clearer...
(09:25:33 AM) eurie: maybe we should float an email to the GO list

(09:25:35 AM) rama: I could be in the minority here...but I wouldn't do push the counts to production just because somebody asked for it...
(09:25:50 AM) gwg: I agree with Eurie
(09:25:55 AM) eurie: and see if people want the counts and why
(09:26:04 AM) gwg: and also ask about the column ordering

(09:26:07 AM) rama: Yeap, good idea Eurie
(09:26:22 AM) eurie: and say we're having display issues in order to deal with (1) accurate numbers and (2) data presentation.
(09:26:41 AM) eurie: no, i think we should leave the interface opinion off
(09:26:49 AM) eurie: we're asking the GO list about functionality

(09:27:03 AM) eurie: not about specific interface design at this point
(09:27:10 AM) rama: i agree with eurie
(09:27:11 AM) gwg: no, but it was the column ordering that brought this whole issue up!
(09:27:32 AM) eurie: yes, i agree, because it's a tangible visible issue that people can complain about

(09:27:42 AM) eurie: we ask them about the data in return
(09:28:05 AM) eurie: what is they want to be able to do with the information?
(09:29:35 AM) gwg: ok then
(09:30:13 AM) rama: we are running out of time..

(09:30:17 AM) sjcarbon: action item: ask GO list for clarification
(09:30:25 AM) rama: yes, thank you Seth
(09:30:30 AM) sjcarbon: can i give this to you, gwg?
(09:30:55 AM) gwg: yup, I'll do this

(09:30:58 AM) rama: Do we have a release date in mind?
(09:31:17 AM) sjcarbon: i think that this is the only think left.
(09:31:19 AM) ben: did we decide 9006 or 9012, or throw to go list?
(09:31:27 AM) sjcarbon: i've already tagged an RC1

(09:31:27 AM) sjcarbon: .
(09:31:43 AM) rama: we haven't decided Ben
(09:31:50 AM) sjcarbon: we'll ask the GO list.
(09:32:29 AM) rama: coming back to release date?

(09:32:45 AM) gwg: when we have this issue sorted?
(09:33:03 AM) sjcarbon: i think that all of the show-stoppers are taken care of. as soon as this is resolved, we can release.
(09:33:25 AM) rama: okay.
(09:33:29 AM) gwg: YES!

(09:33:32 AM) ben: should I put RC1 on go-dev server?
(09:34:21 AM) rama: we are not testing now...I don't think it matters Ben
(09:34:28 AM) sjcarbon: if it's for testing, but i think that we'll probably do all of the fiddling with toy for now.
(09:34:50 AM) sjcarbon: ok, we're at an hour and change--let's wrap up and continue on email.

(09:34:50 AM) ben: ok
(09:34:59 AM) rama: okay
(09:35:00 AM) rama: bye
(09:35:01 AM) sjcarbon: i'll send out minutes and action items from the wiki.

(09:35:01 AM) ben: here is your historical reference for the day:
(09:35:02 AM) ben: I am also of the opinion that qualifiers should be destroyed.
(09:35:04 AM) sjcarbon: thanks!
(09:35:15 AM) sjcarbon: (for coming to the meeting)

(09:35:56 AM) sjcarbon: bye!
(09:36:28 AM) gwg: see y'all around
(09:36:34 AM) gwg left the room.
(09:36:37 AM) pascale: bye

(09:36:40 AM) pascale left the room (quit: Quit).
(09:38:08 AM) rama left the room (quit: Quit: Quitting!).