OE IRC 22Mar07

From GO Wiki
Jump to: navigation, search
area51.slashnet.org has activated the following mode : No External Message
area51.slashnet.org has activated the following mode : Only Operator Change Topic
4:25 PM
jclark has joined the channel
pfey has joined the channel
4:30 PM
tberardi has joined the channel
tberardi: Hello!
midori: hi
tberardi: Anybody there?
jrichter has joined the channel
jclark: yup
jrichter: Hi everyone.
jclark: hi
hjd_mgi has joined the channel
hjd_mgi: hi all
hjd_mgi: Alex may not make it today; we have Journal club overlapping
4:35 PM
jrichter: Journal club?
jclark: hi Harold, was friday successful?
midori: Jane's at a workshop in Manchester.
MelissaH has joined the channel
MelissaH: hey there
jrichter: Hi, Mel.
midori: Is anyone else coming, or should we get started?
jrichter: I don't know of anyone.
tberardi has quit the server saying: Ping timeout
midori: Then let's GO! ... Tho I don't recall seeing any responses to my plea for topics, so it's back to the old "How is 1.1 coming along?"
jrichter: It's okay. I've got quite a bit of news to discuss.
midori: Well, then, don't let me inhibit you.
jrichter: It pertains to the 1.1 question...
jrichter: First, my funding has changed.
jrichter: I'm now 60/40 funded on GO/NCBO.
4:40 PM
jrichter: This actually has very little effect. My NCBO funding ALSO goes to OBO-Edit development, because some NCBO projects (like Phenote) use OBO-Edit as a development toolkit.
midori breathes a sigh of relief
jrichter: But, NCBO has really different priorities. Phenote needs all kinds of experimental/advanced features, so for the last week and a half I've been focusing all my attention on OBO-Edit 2.0.
jclark: what kind of features?
jrichter: Well, for example, last week I spent 3 days working on graph visualization stuff.
midori: Ah ... we just had Michael Montegut visiting (for those of you who don't know, he's the NCBO project manager), and he was very interested in the prospect of graph-based editing.
jrichter: In 2 weeks, I'm expected to have a working Graph Visualization widget available that can be plugged into OBO-Edit and Phenote.
jrichter: And I think it'll work.
tberardi has joined the channel
MelissaH: cool!
jrichter: I built a new component called the "Graph Sandbox" that I think you'll like.
jclark: great!
jrichter: You drag a bunch of terms into the sandbox, and then OBO-Edit uses the reasoner to show what (possibly implied) relationships exist between those terms.
MelissaH: what are the differences between your tool and those at the bioportal and Ontology lookup (EBI)?
midori: that sounds like it has great potential to help us clean up GO relationships ...
MelissaH: reasoner power, yes?
jrichter: I'm not familiar with those tools in any detail, Melissa, but some differences would be editability, graph interactivity, and reasoning.
jrichter: Plus my tool will have some really useful search capabilities.
midori: EBI OLS doesn't do squat with relationships, for one thing ...
jrichter: But I'm ALSO working on something entirely different...
4:45 PM
MelissaH: of course!
midori: [cue Python quotes]
jrichter: OBO format 1.3 will include a gaggle of new features, including term post-composition and the ability to store annotations in OBO format.
jrichter: So this week, I've been working on that.
jrichter: Annotations are stored using a new [Annotation] stanza.
jrichter: The [Annotation] stanza is actually implemented in terms of regular OBO instances, but there's a lot of messiness involved in correctly representing annotations as instances, so the [Annotation] stanza provides a succint shorthand. When the annotations are loaded, OBO-Edit silently transforms them into messy instances behind the scenes.
jclark: outreach group applaudes
jrichter: Term post-composition (also sometimes called pre-composition) is a way of creating anonymous cross product terms.
midori: Methinks I'll understand that [annotations] a bit better when I see some examples.
jrichter: The good thing about the [Annotation] stanza is that we can use it to represent all kinds of annotations. Go annotations, Phenote annotations, and almost anything else.
jrichter: Anyway...
jclark: anything that captures annotations easily is fine by me
MelissaH: so what about dumping annotations into oboedit/phenote, will they have to be in this format or will it accept different formats and turn them into obo format?
4:50 PM
jrichter: The latter, as long as someone writes an adapter for the other format.
jrichter: Let's say that you want to annotate something to the term "mouse heart development during embryonic_stage_3" (or whatever), but that term doesn't exist. Rather than having to create a new term with an id, you can use a special shorthand that looks like this:
MelissaH: wonderful!
midori seconds Mel
jrichter: development^involves(mouse_heart)^during_stage(embryonic_stage_3)
jclark: great!
jrichter: (But replace the term names with their identifiers)
midori: Lots of annotators need that sort of thing, yesterday.
jrichter: Anyway, I'm two thirds done with the annotation and post-comp stuff. I expect to be finished by Friday night.
jrichter: The upshot of the split time thing is that I plan to spend 3 days a week on OBO-Edit 2.0 stuff, and 2 days a week on OBO-Edit 1.1 bug fixes. (I'm giving more time to NCBO right now because I already owe NCBO some time
jrichter: )
jrichter: Okay. Now I'm done.
Alex_MGI has joined the channel
midori: All sounds good; we're looking forward to getting our grubby little paws on OBO-Edit 2.0.
jrichter: Wait one other thing: In case you missed it, Bioinformatics has officially accepted our final draft of the paper, so we're good to go.
4:55 PM
jrichter: There will be an OBO-Edit 2.0 alpha release within 3 weeks.
hjd_mgi: This sounds very interesting. However, for instance, David H is wondering when 1.1 is going out, specifically because he wants to stop having to "de-Dos" his files passing them to a unix system
MelissaH: so john, now that you are not visiting zfin soon, are there any other meetings that you might be able to do some user testing at?
hjd_mgi: Can we have some sort of cut-off on the 1.1 addtions, etc? 
jrichter: OBO-Edit 1.1 additions are cut off. As soon as you guys stop finding bugs, it's released. Even the User's Guide is up to date.
midori: I was just about to send "How are things with OBO-Edit 1.1 betas?" ... and Harold beat me to it.
jrichter: Mel - not for a while. I'm definitely not going anywhere before May.
jrichter: Is anyone else having Midori's unhandled exceptions problem?
midori: I did find that bug in the verification plugin, and it's starting to annoy me (can't remove a nonsense check that I added specifically for bug investigation ... grrr).
midori: I'm not even having the exceptions  problem reproducibly myself! Maybe we shouldn't let it hold up release, unless it does affect anyone else.
jclark: John can watch our editing meetings any time on webex without travelling. 
midori: Good plan! The next one is tomorrow, and there's another one on Monday.
MelissaH: yeah thats a great idea. John, can I have you watch us do some editing sometime? How do I get webex-ey?
jclark: true
jrichter: Midori - If you delete the ~/.oboedit/verify.xml file, it should reboot your Verification plugin editing.
midori: John, shall I forward the webex invites to you?
5:00 PM
jclark: john can also set up his own webex meetings for melissa
jrichter: I'll watch you edit tomorrow, but only if I can crouch in the bushes while I do it, and mumble creepy things like "Oh yeah, you like to click the edit menu, don't you."
jclark: that's creepy
jrichter: So go ahead and forward the webex invite. I'll get my trenchcoat!
tberardi is off to meeting, catch you later
MelissaH: I am laughing. I can only imagine all the horrible things you'll see us doing.
tberardi has quit the server saying: Quit: 
midori: Re verify.xml: I haven't tried simply deleting it, so will do; but I did try replacing it with one that had my cecks (and only the ones I wanted) and it didn't "take". I'll let you know how it goes ...
midori: (er, my checks ...)
jrichter: If that doesn't work, look for an xml file that has the same name as your check and delete that.
jrichter: I can't remember off the top of my head how those custom checks are stored.
midori: definitely haven't seen any of those so far; I do see relevant tags & values within verify.xml
jrichter: Make sure OBO-Edit isn't running when you delete the file. Otherwise OBO-Edit might try to rebuild the file when it shuts down.
midori: ok
jrichter: Do we have anything else to say about OBO-Edit 1.1? Otherwise, I had a question for the group.
midori: I just have one 1.1 question: who else has tested what? Everybody happy with it?
5:05 PM
jrichter: Good question.
jclark: I tested a fair bit and got on much better this time
hjd_mgi: Anything I did worked
MelissaH: me too
hjd_mgi: but my needs are modest
jclark: I can't get to the bugs page right now
jrichter: Hey, Midori... did your unhandled exceptions bug start appearing around the time you added the nonsense verification check?
midori: sorry, nope, the nonsense check and unhandled exceptions are actually plaguing me on different machines (the former at work and the latter at home).
jclark: got there. I tested a lot and have no outstanding bugs
jrichter: poopy
Alex_MGI: It now runs again for me (beta22), no java version problem.
jrichter: Good. Then the bug tracker items probably really are the only outstanding bugs.
5:10 PM
jrichter: Let me move on to my question, then...
jrichter: Could everyone tell me the 2 or 3 most common types of searches you do in OBO-Edit? I'm trying to come up with a way to making searching quicker and easier on OBO-Edit 2.0.
jrichter: please change "to making searching" to "of making searches"
jclark: tell you now?
jrichter: Yeah. If I asked over email, only you and Midori would respond. Here, everyone's trapped.
jclark: aha!
midori laughs out loud ...
jclark: I usually do compound searches on name 
jclark: and ancestry
midori: I search for a single ID all the time, but I know I am eccentric in this respect.
MelissaH: I usually search name and synonym.
MelissaH: boring
jclark: Midori and I have the two ends of the spectrum covered
hjd_mgi: Most frequent search for me is words contained  a term or synonym, since I use it for annotation. Second is sometimes adding an "AND"  ; Third is specific ID 
jclark: I don't remeberany ids
midori: (well, not literally *all* the time. but lots more than anyone else)
Alex_MGI: I do mostly simply term name or ID searches.
jclark: I like the compound search a lot and am very happy with it
jrichter: It sounds like everyone would benefit from the following new feature:
MelissaH: I want to search for terms with more than one is_a parent.
MelissaH: (sorry)
5:15 PM
midori: That said, I've constructed a compound filter to get a term, its siblings, and their ancestors. Just for fun (and to show off).
jrichter: Press Ctrl+F and a search box pops up where you can type in search terms separated by AND and ORs.
jclark: can there be brackets?
Alex_MGI: Wow, you should get a software patent.
MelissaH: ooohhhh.....
midori: aren't those evil (software patents, not brackets)?
jclark: evil brackets would be great!
Alex_MGI: Yeah, but the USPO would probably grant it.
jclark: doing and and or things sounds hard
jrichter: Jen, you bet. The really tough guys could also put in complex expressions, like "(ancestor has name 'lenny' or 'carl') AND kinase"
jclark: could it be nested or something?
midori: aside: The Aussie patent office granted someone a patent for the wheel.
jclark: it does sound quicker once you get the knack of it
jclark: All that clicking takes time when I make a filter
Alex_MGI: Despite my humor, I think this is a great idea.
jclark: would you leave the normal way too?
jrichter: responding to Midori's aside: That sounds like a lefty plot to dismantle patents all together. Like when a liberal lefty congressman occasionally calls for instituting the draft to scare everybody.
jrichter: Jen - of course.
jclark: could you put in the auto indent featur that we have on xml so it auto indents?
midori: Anyway, I also think there's a lot of room to make link filtering more intuitive.
jrichter: Jen - this would probably be on a single line.
jclark: link filitering is really hard
jrichter: Any suggestions on how to improve this?
jclark: for complex ones I think it is better to have them nested
jclark: I have no idea how link filtering works
5:20 PM
jrichter: Link filtering provides a way to only show links that match certain criteria.
jrichter: Normally, you 
jrichter: design a link filter to only show links of a certain type, or that belong to a certain namespace
Alex_MGI: I think the point of using the CTRL-F is to allow a quick way of doing fairly simple searches.
midori: I can't tell how to combine a link filter with a term filter (if it's even possible), and neither can Jane. She put in a feature request, SF 1604421 (https://sourceforge.net/tracker/index.php?func=detail&aid=1604421&group_id=36855&atid=418260).
jclark: I think that simplifying link filters would be good
jclark: alex, I see you point
jrichter: Midori - when you click the "filter" button, both your link filters and term filters are applied.
jclark: most of my fliters have two or three criteria
jclark: would it be possibel to make the language of the link filter more intuitive?
jclark: Maybe link filtering just is unintuitive
jrichter: Maybe what I should do is remove the "Parent" and "Child" aspects of the link filter. Almost no one seems to use them, and I think it complicates things for most people.
midori: We do tend to think of link filtering mainly as a way to distinguish is_a frorm part_of in a search ...
jclark: is it so we can do things like search for terms with 'bob' in the def that have an is_a ancestor 'development'?
jrichter: No.
jrichter: Link filtering isn't terribly useful for searching, unless you're trying to find a particular relationship.
jrichter: It's mostly useful for actual filtering, where you want to hide particular relationships.
jclark: oh I see
midori: Ah. What should we do, then, to narrow a search by relationship type?
jclark: but it's different from the flower button isn't it?
jrichter: Melissa uses this feature all the time, because she doesn't want to see non-transitive relationships in her ontology.
jrichter: Midori - you can't do that. I'm not really sure what that means, actually.
5:25 PM
jrichter: Jen - the flower button IS a specific kind of link filter, with some additional interaction with the reasoner.
MelissaH: I put it in my global filter so I never think about it.
jclark: maybe if you watch us edit for a while then you'll see lots of optimizations that we could be making in our searches
jclark: it's hard for us to know what we're mssing
midori: What we mean is, how do we build a filter that gets things like the 'is_a children named bob' that Jen described (or something like that ...)?
jrichter: Midori - "is_a children" of what?
jrichter: Of the currently selected term, or something?
midori: Jane' s example was "show me all the terms that have a part_of parent that includes the word 'metabolism'" -- does that make it clearer?
jrichter: Aha. That is not currently possible, but it will be in OBO-Edit 2.0, because I'm going to add the intimidatingly named feature "parameterized search aspects"
jrichter: Right now, the best you can do is: [Ancestor] [Name] [contains] "parent"
jrichter: But that will give you ALL ancestors, not just part_of ancestors.
jclark: that search would be useful
midori: Well, that takes care of my jargon quotient for the day. (The new search capability will be cool too.)
jrichter: When we have parameterized aspects, you'll be able to say: [Ancestor(part_of)] [Name] [contains] "parent"
jclark: cool!
jclark: is there a shorter name?
jclark: that's going to be really hard to discuss
jrichter: I wouldn't worry too much about the name. I think no one in the world will need to have a name for it but me. For you, it'll just be the box next to the "Ancestor" pulldown.
5:30 PM
jclark: great!
jrichter: Just about time to go...
jclark: should we all start editing in 1.1 this week for intensive testing?
jrichter: Certainly.
Alex_MGI: So I saw a seminar today where the screenshots were still of DAG-Edit.  I wonder when people will notice OBO-Edit is better.
jclark: that's really a long time
jrichter: Maybe I need to add some messages to the DAG-Edit download page saying that this tool is deprecated.
Alex_MGI: probably a good idea.
jrichter: Actually, I think I'll just take down the DAG-Edit page.
MelissaH: and add a redirect.
jclark: did you see erika's bug?
midori: oooh, tough love!
jclark: Ijust rememebred
jclark: she needs to load go files and the editor crashes
jrichter: Jen - yes. There's a 95% chance that she hasn't configured her flat file settings correctly. The file she's loading is so old, she needs to do some really specific things to get it to load right.
jclark: cool
jrichter: It actually crashes, or it doesn't load the file?
jclark: did you tell her what?
jrichter: (Sometimes people say "crash" when they mean "doesn't work right")
jrichter: I'm in the midst of posting that advice on the bug report page right now.
jclark: great 
jclark: ta
jrichter: And if anyone has any students that need a project, please send them my way. I have tons of stuff for little compsci folks to do.
5:35 PM
jclark: I haven't found any oboedit programming thing that I can do yet
jclark: except to rename the search button
jrichter: Are you asking for a task, Jen?
jclark: do you have any really really easy ones?
jclark: yes
midori: Off on a tangent re patent humor: http://improbable.com/ig-pastwinners.html#ig2001; look for 'technology'.
jrichter: Jen - it's fairly easy to extend the Extended Info Plugin, and there's all kinds of different analyses you could add. Perhaps you could think of a new stat you'd like to see in the Extended Info Plugin, and we'll start an email thread about how to add that.
jrichter: And you'll gain tons of experience working with the datamodel.
jclark: ok
jclark: I'll try to think of one.
jrichter: Ooh! And maybe you could keep a little developers blog about your work!
jclark: ok 
jclark: does anybody want any addition to the extendedinforplugin?
jclark: while we're here?
MelissaH: mmm I had a request for one a while back, I'll take a look and remind myself.
MelissaH: and you.
jrichter: (If you can't think of anything, one idea might be to add info about how many terms are in each namespace)
jrichter: (Or how many obsolete terms have no replacement info)
MelissaH: or how many terms with no parent of a particular type.
MelissaH: how many is_a or develops_from incomplete terms are there?
jrichter: That's a good one!
jrichter: Maybe start with IS_A, because it's guaranteed to exist in every ontology.
5:40 PM
MelissaH: certainly!
jclark: shall I do that then?
jrichter: Yeah, that's a really good idea, and it would be useful.
jclark: it may take years.
jclark: this is my confusing weekend project
jrichter: I predict it'll take < 3 days.
jclark: okey dokey
jrichter: Stick around in the chat after everyone leaves, and I'll get you started.
jclark: ta 
jrichter: Bye everyone else!
jclark: bye
Alex_MGI: Bye.
Alex_MGI has quit the server saying: Quit: Quitting!