From GO Public
|Line 358:||Line 358:|
<span style="font-weight: bold; color: blue">gwg</span>: so see you all then!
<span style="font-weight: bold; color: blue">gwg</span>: so see you all then!
Latest revision as of 17:58, 13 February 2007
gwg: OK, let's go then. Ben is first up with the report on amigo.geneontology.org
hitz: OK, stuart is having some trouble with some maintenance db copying scripts
hitz: he would like until next tuesday to get it all squared away.
hitz: I don't like to anhoy him because he's the system admin.
hitz: any problems?
gwg: I don't think so
hitz: good, next item
gwg: GO loading refactor item
hitz: just a note that I am writing specs for speeding up the go loading. On page, if this kind of thing is interesting to you.
gwg: the link is on the wiki
hitz: thx amelia, typing too fast
gwg: excellent! next item is the requests for the June AmiGO release
gwg: working groupers, have you all been wracking your brains to think of things to ask for?
hitz: there is some disucssion in the talk section, but I find (and maybe Chris will agree/disagree) that the talk pages are not that cool. Better to discuss by email
gwg: ok, cool Ben.
gwg: have people had a look at the feature requests tracker to see what's in there and if there's anything they want to add?
cjm: Ben: I agree. It's
susan has joined the channel
gwg: AmiGO request tracker url: https://sourceforge.net/tracker/?func=browse&group_id=36855&atid=494390
gwg: hi susan, we are just talking about the new feature requests for amigo
gwg: well, I'm talking mainly
susan: ok sorry I'm late
gwg: so there are no new requests from anyone?
gwg: could you canvass your colleagues and see if they have any requests?
eurie: Are there any requests already on the tracker that you really really want?
donghui: will do
gwg: at the moment, there are a number of little fixes, and then the main concerns are sorting results
donghui: I want that feature too
gwg: and a few modifications to the blast search
gwg: from what has come in on the mailing list recently, it sounds as if data downloads is also something that users want
cjm: I think people want more data download options, but . this will require a couple of additions to go-perl that Seth and I are working on. We can put in a request for this as a placeholder
susan: I agree
gwg: (the gohelp mailing list that is)
rama: How about searching by kingdoms?
gwg: I think that the best course of action might be for the AWGers to see if they can think of any further feature requests, or see if their colleagues have any, and then we can prioritize what will be done for the June release
hitz: that one is trickier because it requires change to db schema
hitz: (re: search by kingdom)
sjcarbon: species is already in the database--couldn't we just have kingdom groups?
gwg: perhaps the next AWG meeting in two weeks time could be used for prioritization
rama: okay, is there a plan to do it later than June
cjm: seth: people may want kingdom, order, family, phylum, genus, ...
sjcarbon: fair enough
pascale: did we not discuss using the NCBI structure to do this?
pascale: is that possible?
gwg: I think the species question is definitely an AmiGO 2.0 thing as we'll have to integrate a taxon tree into AmiGO
hitz: I thnk the NCBI taxonomy file is parsed to get the species file, but I think we should leave this to 2.0
cjm: Changes to the schema aren't so hard - there is a standard way of storing a taxonomy tree in a SQL database. There would also be places in the API that would need hooks added, this is the bulk of the work
cjm: Ben: the nodes but not the hierarchy is loaded
eurie: Pascale, why don't you put that in the tracker?
gwg: we've already got something in the tracker along those lines, I think...
gwg: there's definitely some rambling on the AmiGO big ideas page about it
cjm: (We can also generate an OBO file from the taxonomy and treat it just like an ontology. This has a lot of advantages. may be less efficient though, and would slow down loading)
rama: yes, we have it in the tracker already
eurie: got it - I had the filter set to "AmiGO 1.0" so only saw a subset of items.
hitz: chris - I agree its not hard, but I think it's an added complication given a load re-write. Ask me again in april
gwg: OK, so action item from that discussion: everyone to think of AmiGO feature requests before the next session, and add them to SourceForge. We'll prioritize what should be done at the next meeting
gwg: Seth, do you want to mention the work you've done? Is it ready for showing to the AWG, or do you want to leave it a while?
sjcarbon: some things are showable, some maybe should wait
gwg: the term finder?
sjcarbon: the termfinder kinda needs another component, which is what I completed
sjcarbon: a kind of file manager
sjcarbon: it will hook in to the slim and termfinder
sjcarbon: (at this point, the slim client is closer to completion than the termfinder)
gwg: slim as in map2slim?
sjcarbon: something will be working with that later today or tomorrow
gwg: OK, great. If you send the URL out, we can all try to find some bugs to keep you busy
eurie: how about the auto-complete options?
gwg: how about the autocomplete demo?
donghui: there is a demo already? great
gwg: Chris, Ben - we're just about done on the discussion front if you need to scarper
hitz: bye all
hitz has quit the server saying: Quit: Quitting!
sjcarbon: sorry-phone call for chris
sjcarbon: gully wants to talk
gwg: so is the autocomplete demo ready for action or are you still working on it?
sjcarbon: is the autocomplete demo
gwg: do you want people to have a go and tell you which they prefer?
sjcarbon: i'd love that (sorry, half of my attention is trying to figure out why the demos worked last night when I left and are silent now)
val: working for me, cool
rama: Seth, you should post some instructions on how to test/what to expect...
rama: yeah, it is working for me too.
sjcarbon: the first seven and the last sevrn use the same algorithm, but are tied to a different set of widgets
gwg: Do you want to email everyone with the URL and some instructions? The people who aren't here can also comment then
gwg: Does anyone have any other business?
gwg: (AmiGO related business, that is)
gwg: going once
gwg: going twice...
sjcarbon: i think eurie won
eurie: but I think you were louder, seth. =) do we have a time frame for the autocomplete comments?
gwg: comments to come in before the next AWG meeting
gwg: presumably the sooner the better, yes?
sjcarbon: it gets complicated; i'd like to have a discussion sometime about widgets and the like
pascale: can you explain what the number means on top of the list?
sjcarbon: as well as the algorithms
sjcarbon: that's the number of things returned
pascale: really? you only see the first ten or so then?
pascale: ok thanks
sjcarbon: also, the file manager is working again (i was having NFS problems)
rama: Please also explain what the 1,2,3,4 pull downs mean in each widget
rama: all in the email is fine.
gwg: seth, do you want to email about scheduling the discussion about widgets, algorithms and similar things? Will it be an AWG thing?
sjcarbon: just an id so we know which we're talking about
sjcarbon: the ones that start with a G are doing gene products, what the ones wothout are doing terms
donghui: what's the 'process' button for?
sjcarbon: something fun to press
donghui: i thought it should be 'auto'
sjcarbon: i'll remove it to avoid confusion
eurie: we have some time now - should we do some discussion now?
sjcarbon: is a file manger for programs like map2slim
rama: eurie, you mean on autocomplete
eurie: for whatever seth wants to discuss
rama: Seth, your call
sjcarbon: is there other business?
gwg: no, I don't think so
gwg: nothing else on the agenda, anyway
eurie: me? no, my wait was only for the time frame for comments
sjcarbon: well, let's see what I can explain in ten minutes
sjcarbon: for a AmiGO2, it's becoming obvious that widgets (buttons, sortable tables, etc.) will be a time sink
sjcarbon: so we have the option of using a third-party widget set or going alone
sjcarbon: third-party will give us speed of application turn-araound, but decrease flexibility
sjcarbon: for example, in the autocomplete
sjcarbon: the top seven lines are from a third party
sjcarbon: and there is no real way to separate out the number of returned items from the actual returned tiems
sjcarbon: whereas the bottom severn are something i put together in an afternoon
sjcarbon: but never got around to completing (doesn't listen to mouse yet) because I ran out of time
sjcarbon: the issue in a nutshell
sjcarbon: until now, I've been rolling my own (sortable table)
sjcarbon: but there is definitely an argument for just grabbing someone else's
gwg: especially for stuff like tablesorting... there are a million sortable tables out there
sjcarbon: very few *fast* ones though
sjcarbon: which is why I went my own way in that case
donghui: autocomplete: sorry to interupt. i suppose the results are sorted by alphabetical order. If I am searching for a term 'kinase', I would like to see that term listed on the very top; followed by other terms. Is this reasonable?
gwg: I think someone else mentioned having the autocomplete results ranked by 'relevance'
sjcarbon: that would be algorithm "G2" i believe
susan: I'm confused isn't that a gene product one?
sjcarbon: yes, it is
val: will the results eventually be scrollabe?
sjcarbon: for 5000? i think that's a bad idea
sjcarbon: a lot of data to transfer
sjcarbon: if the autocomplete isn't quick, it probably will be more of an problem than anything else
gwg: and if your search query is vague to start with, it probably won't help
susan: with the second one how do you select something from the list? e.g. type methylation and you want to choose DNA methylation from the list - do you have to type the missing bit?
sjcarbon: for the second set (the second seven)?
susan: BBOP rather than Dojo
sjcarbon: up and down arrows and then tab or return
susan: doesn't seem to work for me
val: I see... i"m typing things like actin though...325 results and the most relevent aren't first. Perhaps there are other ways like prioritizing term banes above synonyms etc
val: yes BBOp is better
sjcarbon: sorry--this seems to be running pretty slow hammering on my little toy server
sjcarbon: susan: make sure that the cursor is flashing in the text box
gwg: perhaps people should test it over the next few days so they can get a feel for how the response would be when the entire AWG wasn't testing it!
susan: ahh got it - thanks
gwg: Shall we draw proceedings to a close? Our hour is pretty much up
gwg: I'll put the action items and the transcript on the wiki
rama: great. Thanks Amelia for chairing the session
gwg: Next meeting is in two weeks time, the 27th
gwg: so see you all then!