OE IRC 4May06
jrichter: While we wait, has everyone gotten a chance to download the new docs I posted last night?
jrichter: They're on sourceforge in the beta18 release section.
kchris: Just did it now, so am just starting to look at them.
midori: yep, got 'em, and have skimmed a bit
jrichter: Looks like it's time to get started.
jrichter: Midori, did you have any agenda items for this meeting?
midori: sort of, in a non-commital kind of way ...
jrichter: Maybe we should do 'em!
midori: 1. anything interesting from your Berzerkeley visit?
jrichter: Not in terms of actual OBO-Edit development. However, Suzi has some ideas about things we should work on in the future...
jrichter: This visit was devised so that I could meet Mike Bada, a guy at the University of Colorado Health Sciences Center who is writing a tool to infer links in the GO from term names.
midori: OK; wanna list 'em now or postpone (I have no pref)?
Harold: howdie folks
jrichter: He and I are going to work together to design an API for his tool. That API would eventually get integrated into OBO-Edit.
midori: oh yeah, I've met Mike ... seems like a nice guy
jrichter: Go ahead and list 'em.
midori: oh .. I meant list S's new feature ideas ... but agenda items are:
midori: 2. User guide progress update
midori: 3. testing beta18 - anything to report?
midori: 4. filtering review
midori: that's all I came up with ...
J-Lo: can we also talk about building obo-edit from cvs?
jrichter: That sounds like a good one. Let's make that item 5.
midori: ok by me
jrichter: Just to finish up on item 1.
jrichter: It's possible that I'll be sidetracked in a few weeks to focus on his project. I may also be pulled toward OBO/OWL integration in the next version, even though that's not a high working-group priority.
jrichter: A lot of the research types are working in Protege, and it would be nice if we could use their work without a lot of gymnastics.
jrichter: So that's the end of my Berkeley report.
midori: well, just keep us informed. If we can get OBO-Edit released before you get diverted, so much the better
jrichter: Ready for the User's Guide update?
jrichter: I WILL NOT be diverted before the OBO-Edit release.
jrichter: Here's what's still missing:
jrichter: The Parent Plugin
jrichter: The entire History Section
jrichter: The Complex Class Plugin
jrichter: The Configuration Plugin
jrichter: 1/2 of customizing the interface
jrichter: The Graph Viewer Plugin
jrichter: Introduction to Plugins
jrichter: That's it.
midori: getting there!
Harold: I turned in graph viewer plugin, history plugin, and id generator plugin
gwg: do you still have a job, then?
jrichter: I do, Amelia. Remember, I was granted a reprive because I was sidetracked by this visit.
jrichter: I get an extra 4 working days.
jrichter: Thanks to everyone who contributed, by the way!
midori: Ready to move on?
kchris: So, for those of us whose assigned sections have been completed, our task is just to proof them and give you comments?
jrichter: If you contributed a section and I haven't integrated it, I'm probably building it up to the exhaustive style in the rest of the guide. That is, every single friggin button is documented.
jrichter: That's right Karen.
jrichter: Oh, that reminds me...
Alex_MGI: You've made excellent progress John -- Thanks.
J-Lo: Oh - I've got this how to make a GO slim using OBO-Edit walkthrough too - will i just stick it in SF?
jrichter: Jane - I think we ought to integrate that into the User's Guide. I'll add a how-to section at the end.
midori: might as well put it in SF
J-Lo: will do
jrichter: When you say "put it in SF", do you mean commit to cvs?
J-Lo: not sure what I meant actually
J-Lo: shall i put it in cvs?
jrichter: If you could integrate it into the documentation part of the cvs, I could put it in the user guide, and it will get posted online with the rest of the docs.
jrichter: I still need to talk to Mike about how to make this happen.
jrichter: The thing that I was reminded of - I'd like some volunteers to go over the "Viewing Ontologies" section. Melissa had some great comments, but I disagreed with several of them. I'd like another reader.
kchris: He (MIke) is back from his trip now to be able to talk to him
kchris: I'll read it.
jrichter: K - I'll email him today, if possible.
carol: I could look it over if you like, as well.
jrichter: Thanks! We're looking to make sure it's logically ordered, and that the following point is clear-
jrichter: The sub-selection is a single TERM. The main selection is a collection of RELATIONSHIPS.
jrichter: Did anyone have anything else to say about the User's Guide before we move on?
J-Lo: do i have sf cvs write access?
jrichter: I think you do. If you don't, I believe you have the power to grant yourself CVS access.
jrichter: The docs are at go-dev/java/oboedit/docs/
jrichter: Midori - should we move on?
J-Lo: Midori's on the phone
jrichter: (Forgot to say - An introduction to OBO Ontologies is also incomplete)
jrichter: Okay. Midori's agenda item 3 was "testing beta18
jrichter: ". Any thoughts?
jclark: I haven't done any b18 testing because I was a bit nervous about removing escape characters manually.
Harold: same here
jrichter: Actually, I don't understand why that's necessary... beta18 doesn't have that problem.
jclark: it has a new problem
midori: OK, I'm back
jclark: I can't remember the exact details but it puts escape characters in that I think mean that oboedit16 can't load the file
jrichter: What problem is that?
jclark: Midori suggested that we remove them with a search and replace
jrichter: Has this bug been on sourceforge and closed?
midori: that's correct - beta18 can deal with the escape characters that it puts in (unlike beta17)
jclark: I sent it to the list and then wrote to you to ask if you'd got it or if you wanted me to put it on sourceforge but you didn't reply
jclark: I figured you were budy and you'd get back to me
midori: it's not exactly a bug; it's more an incompatibility between 16 and 18, which bites us because we're officially using beta16 for files to commit in gocvs
jrichter: Hmm... that might have been during one of our recent email outages. But I always want bugs on the sourceforge tracker. I use it as my to do list every morning.
jclark: i'll add it
jrichter: Is this really a beta18 bug, or a beta16 bug?
J-Lo: shall we just move to beta 18 as our commit version - that solves this doesn't it?
jrichter: It does IF beta18 works and is generating correct OBO output.
kchris: Probably depends on whether you want escape characters in the file or not, as to whether it's a b16 or b18 bug...
Harold: If everything else works
jclark: we've not tested it at all
jclark: I think we should do testing first
jrichter: I know what to do!
Alex_MGI: I think we better wait until we happy with the testing.
kchris: I agree
midori: there will also be a big diff when we switch to beta18, because it puts dbxrefs in a different order
jclark: what's your plan john?
jrichter: I'll build a little utility that will convert beta18-OBO to beta16-OBO. Then we can test beta18 and submit your work to CVS in a version others can use.
jrichter: If it's just an escape character thing, it should be easy to build a converter.
jrichter: (Like 15 minutes easy)
jclark: wouldn't that just be a search and replace? We could do that ourselves
jclark: if t's just removing all the escape characters
jrichter: It probably is more like removing escape characters in certain contexts.
jrichter: That might be easier for me. I'm going to install both versions as soon as possible to see exactly what's going on.
jclark: or could you just fix the bug and rerelease?
gwg: and reordering dbxrefs
jrichter: The dbxrefs have been reordered on purpose, I believe.
midori: Well, I *have* used beta18 for the few little changes I've made lately (by removing escapes and then load 'n' save with beta16), and pretty much all has gone well.
jrichter: They were being sorted by description rather than by dbxref.
jrichter: Here's the problem with just doing a fix and releasing -
kchris: What was being sorted "by description rather than by dbxref."
jrichter: dbxrefs - they were being sorted by descriptions rather than the actual dbxref data.
midori: Today it's been stalling when it's had a lot of paths to select; I wanted to gather more data before doing a bug report.
jrichter: I'm worried that the problem is that OBO-Edit b16 is actually the culprit and failing to read correct OBO.
jrichter: If that's the trouble, doing a new release won't fix it.
midori: (especially since it didn't have the problem yesterday ...)
jrichter: I'm interested to see that bug report...
jclark: I'll dig it out
jrichter: Anyway, if our real problem is beta16, we may need a conversion utility because we're stuck with beta16 as the standard. I'll do some research and then email a report to the group.
kchris: sounds like a plan
jclark: I just sent the bug report to the obo-edit mailing list
jrichter: Thanks, Jen.
jrichter: Shall we move on?
midori: next up: filtering lesson!
midori: Rama wanted to "save all the 'is_a' children (and is_a grand children and is_a great
midori: grand..) of the term Protein Complex in a separate file."
rama: I was just about to mention that. thanks Midori.
midori: Jane sent some instructions on the requisite filter
rama: I tried Jane's instructions, but I ended up saving all the children, not just is_a
midori: there was an exchange, but the outcome was that it would help very, very much to go over it 'live' during this chat.
midori: so ladies & gentlemen, start your OBO-Edits!
jrichter: Some questions before I start -
jrichter: Are the children of Protein Complex is_a completely connected by is_a links?
jrichter: (Actually, that's the only question)
rama: What do you mean?
midori: John - I *think* so ...
J-Lo: protein complex has no parts
jrichter: Perfect. That's the answer I wanted.
jrichter: I *think* you can get away with doing a filter like this:
jrichter: - [Self] [ID] [equals] <protein complex id>
jrichter: - [Ancestor] [ID] [equals] <protein complex id>
rama: Are we doing SaveAs ?
kchris: back up a step, what needs to be checked in the Advance options?
midori: do we have to have compound filtering on?
jrichter: Let's step back -
jrichter: The best way to filter a file is to test the filter in the interface first.
jrichter: Make sure that your root selection algorithm is set to GREEDY.
jrichter: (That's the little root button under your term editor panel)
jrichter: Now, go to the Advanced tab of the filter control.
jrichter: Make sure that Keyword filtering is unclicked.
jrichter: Make sure that Term Filtering, Link Filtering, and Compound Filtering are clicked.
jrichter: Now create the filter I described above.
jrichter: (You know, actually link filtering should be turned off)
J-Lo: ah - some protein complexes _have_ parts though
J-Lo: so you do need a link filter
jrichter: Then we need to turn on link filtering ALSO, and create the filter:
kchris: I'm still having problems with this step: [Self] [ID] [equals] <protein complex id>
jrichter: [Type] [Self] [ID] [equals] "part_of"
jrichter: [Type] [Self] [ID] [equals] "is_a"
jrichter: That's what I meant.
jrichter: Karen - What's the problem?
jrichter: "<protein complex id>" should be replaced by whatever the id is for that term.
kchris: I couldn't figure out where the 'ID' was coming from. I'd forgotten that there were more options listed above in the scroll window
rama: and then hit search?
jrichter: Once you've got your term and link filters set up, click the "Filter" button.
jrichter: NOT Search.
rama: got it!
rama: I meant hit filter.
rama: Still filtering....
jrichter: It could take a while to apply the filter, since this is a complex one.
jrichter: "Descendant" and "Ancestor" filters are always slow. I'm going to work on an optimization to speed these up once the user's guide is done.
jrichter: Has it worked for anybody?
Harold: I lose everything
jrichter: Are you sure your root selection algorithm is "GREEDY" and you've typed all the ids correctly?
kchris: No, I haven't figured out how to add all the constraints yet, but I can get help from Rama
Harold: I have filter set up go_id =
kchris: Not that we have to go through it right now, but I've had another question, that's kind of related
jrichter: Oh, crap! The id for is_a is not "is_a". It's "OBO_REL:is_a". Adjust your link filter accordingly.
kchris: It used to be easy to say give me all children of a particular term, and now I have no idea how to do it.
rama: okay, will redo the filter.
jrichter: Karen - you can do a search like: [Ancestor] [id] [equals] <parent term id>
jrichter: This will return everything that has an ancestor with the desired id.
Harold: Ok, I'm confused; in one case we use ID for the GO id and the other case we use ID for the relationship.
Harold: SHouldn't we be using "property" for the realtionship?
Alex_MGI: My OBO-Edit is still filtering on is_a's -- taking a long time.
jrichter: You need a decent amount of memory for this. Make sure OBO-Edit hasn't hung with an OutOfMemoryError.
jrichter: Harold - I don't understand your question.
jrichter: The Term filter means "Give me Protein Complex and its descendants"
Alex_MGI: Where would that be reported (on a mac)?
jrichter: The link filter means "Only show is_a links between them"
jrichter: Alex: Run the command "tail ~/.oboedit/stderr"
jrichter: If you see a stack dump, it's a good bet you've got an out of memory situation.
Alex_MGI: tail: /Users/adiehl/.oboedit/stderr: No such file or directory
jrichter: Hmm... hold on.
rama: Filtering is over. I think it did it right. I am going through the tree to make sure i got the is_a's
jrichter: Alex - go ahead and turn up your memory settings and restart.
Alex_MGI: Will try that.
jrichter: When I build the descendant/ancestor optimization it'll use a lot less memory as well.
kchris: Hmmm, my minor search to get all children of a given term, left it with the id for protein complex, only gave me 4 terms, and put my display into something useless where it no longer allowed me to view the whole tree...
jrichter: Karen - what exact search were you doing?
kchris: Apologies, I must leave now. Bill has just returned from a trip and doesn't have keys to the house ...
jrichter: Quickly Karen -
kchris: John, I'll have to work on this later and get back to you.
jrichter: You need to have link filtering turned off for this to work.
kchris: It was
gwg: so enigmatic!
jrichter: We'll have to finish that by email.
jrichter: Bye Karen.
jrichter: Rama - is that working for you?
Harold: I have a major problem hee in that the link filter box (the place you type in the text) does not show up; it's "below" the line with the
Harold: self id equals buttons
jrichter: You may need to adjust your screen layout so that there's a splitpane bar there.
jrichter: While people are still working on this...
rama: Here is what I am seeing- GO:8622 is a part_of child, and this doesn't show up in my filter, but
midori: I had the same problem ... tried to futz with my layout to make the search box taller ... managed to get OE to hang :P
Harold: If I send you my layout, can you add it; I don't know xml
midori: I'll email you the one I just did, off-chat
jrichter: Yeah, Harold. I'll take a look at it.
rama: GO:1536, 1535 which are both part_of children of 1534 show up on my filtered list.
jrichter: Do they show up as roots?
Harold: I can expand the window with the AND, etc. but nothing ese
rama: Sorry, please refresh my mind. How can I figure that out?
jrichter: They would be right at the first level of the display.
jrichter: Right under the "Classes" node.
jrichter: Anyway, we're just about out of time, so here's what I'll do:
rama: Yes, they (1534, 1535, 1536) show up right under classes node.
jrichter: Good, that's the correct behavior.
jrichter: That means that there is no is_a path to protein complex for those terms.
rama: but 1535 and 1536 are part_of children of 1534. why are they showing up and why not 8622?
Harold: ooh; those are ours
Alex_MGI: I believe it worked correctly for me.
midori: naughty MGI!
jrichter: I'm not entirely sure, Rama. I don't have OBO-Edit running now. I'll take a look at it.
rama: Okay, can you email me/the list later?
jrichter: When I get home.
rama: Thanks a lot.
jrichter: Yeah, I'll email the list when I've analyzed this.
jrichter: Important! :
jrichter: SAVE THE FILTER. That way you can use it when you save without having to recreate it.
rama: Ah ha! thanks.
jrichter: Anyone who has problems with a filter should save it and email me the file. That way I can do accurate testing.
jrichter: Quickly, since we're out of time -
jrichter: Jane, did you have something about CVS you wanted to talk about?
J-Lo: yep - I manged to download and complie
J-Lo: compile OE, but got this message when i tried to run it
Alex_MGI: Where do I save the filter?
J-Lo: [jane@wolf oboedit]$ ./oboedit
J-Lo: Starting DAG-Edit at /homes/jane/go-dev/java/oboedit using 512M of memory
J-Lo: Exception in thread "main" java.lang.NoClassDefFoundError: org/geneontology/oboedit/OBOEdit
J-Lo: any ideas?
jrichter: (Alex - Save it anywhere you want)
Alex_MGI: What button do I press?
jrichter: Jane - There's probably a problem with what I've got checked into CVS. I probably forgot to check something in. I'll test this tonight or tomorrow morning and send out a report.
jrichter: Alex - Press the little "Diskette" looking button under the "Filter" button.
J-Lo: cool - thanks.
jrichter: Action items for me:
jrichter: 1) Finish remaining docs
jrichter: 2) Analyze escape character bug in beta18 and report
Alex_MGI: When I first pressed it the dialog box said "open" at the top, so I cancelled. Pressing it subsequently yields no dialog box at all.
jrichter: 3) Email the group about Rama's filter and why certain terms show up where they do.
jrichter: 4) Make sure CVS sources compile
jrichter: Alex - The "Open" text was probably a typo.
rama: the button that looks like a folder and the disk one do the same. They open.
jrichter: If you can't reopen the window once you've cancelled, that may be a bug.
Alex_MGI: I got confused so I asked.
jclark: when I press the disckette it says open but it does savet he filter
jrichter: I don't think both windows open - they both just say "Open".
jclark: yes that's right
jrichter: Could someone submit a bug report about the save filter typo?
Alex_MGI: I will.
jclark: you also have to already have a text file in existence that you can choose
Harold: it's kind of like the save layout; you have to have a file already in existence to save it into
jclark: it doesn't create a file for you
jrichter: That's not true - you can type a new file name into that box.
jclark: can you? ok#
jclark: yes I suppose so. I never thought that
jclark: I was interpreting the open command
jrichter: Whoever has the "Filtering" section of the docs should make sure that is clear in the documentation.
midori: sorry, what box??
rama: Clicking on the disk button doesn't do anything anymore. I will use the SaveAs from the pull down.
jrichter: Save As will not work.
jrichter: Save As just saves the ontology.
jrichter: It looks like we have a bug with the filter save/load buttons.
Alex_MGI: One bug report or two?
rama: what are my options?
jrichter: Two - one for the typo, one for the windows not reopening.
Alex_MGI: The folder "open" button does work after canceling at least.
jrichter: Rama - You'll probably need to restart before you can load or save another filter.
Harold: hmm a, after I used the save layout dialog to save an xml file, now the configuration manager is grayed out
jrichter: Harold - Are you sure that the configuration manager isn't still open, but behind the main window or something?
jrichter: Just looked at the clock - I've got to go. There's a plane to catch and a long BART ride ahead of me.
rama: Okay. Have a safe trip.
Harold: positive; there are no other windows other than the IRC open and the main obo edit
midori: Have a safe trip!
jrichter: Harold - see if this is a reproduceable bug. If it is, please submit a report.
Harold: oak creek again?
gwg: happy tripping, John-boy!
Alex_MGI: Thanks and Bye.
jrichter: I don't know what you're talking about Harold. But bye bye!
Harold: when is our next chat
midori: I'll send the transcript around.
jrichter: Thanks Midori. Let's chat again next Thurs.
jrichter: The docs should be out by then.
midori: ooh, just one week? OK!
jrichter: But we can solidify that by email.
Harold: (sfa to oak creek via bart
gwg: see yas!