Perl 6 - the future is here, just unevenly distributed

IRC log for #opentreeoflife, 2014-03-20

| Channels | #opentreeoflife index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:15 kcranstn joined #opentreeoflife
01:26 jimallman joined #opentreeoflife
03:57 mtholder joined #opentreeoflife
03:59 mtholder Hi jimallman. I did not get around to much work on the treebase import stuff. but i did migrate some of the functions that you added to study_import_options branch of api into peyotl. just makes it easier to test...
05:05 23LAAY0GT joined #opentreeoflife
05:10 jimallman joined #opentreeoflife
11:35 kcranstn joined #opentreeoflife
13:37 towodo joined #opentreeoflife
14:31 jimallman joined #opentreeoflife
14:47 towodo jimallman, testing the neo4j applications is not all that hard… I could step you through it at some point if you like
14:49 jimallman towodo: that would be great, thanks!
14:50 towodo let me know when, we can do it using irc
15:44 blackrim joined #opentreeoflife
15:45 blackrim towodo: you here? figured might be easier to talk about this over irc then email just because we can discuss.
15:45 towodo yes, i'm here
15:46 blackrim so, yes, there are issues with "wrong" and "right" with the trees. the issue, as I see it, is that Laura recognizes that Eumetazoa probably "exists" but her tree has some artifacts due to sampling. so it doesn't reflect the general consensus in the field
15:46 blackrim nevertheless, it is a problem that is for sure
15:46 towodo yes. same as in fungi.
15:46 blackrim right
15:47 blackrim the other bit, one sec someone just came in ...
15:47 towodo it all comes down to, what do you want to tell the world, in your paper?  what claim do you want to stand by?
15:47 towodo ok
15:55 blackrim right. i think non monophyletic eumetazoa would be bad. i think these things don't get highlighted in the original papers so people don't notice (i certainly didn't when I read it)
15:55 blackrim for the methodological issue. I agree. i don't like supertree methods for this reason
15:55 blackrim basically, it is impossible to combine two trees honestly without a bunch of junk (e.g., the reason you don't see many supertrees anymore that aren't heavily curated)
15:56 towodo don't be too strict about 'honestly'.  honesty is just being truthful about what you're doing
15:57 blackrim the reason you can't just combine a bunch of trees blindly is because, first there is what trees you decided would be included (of course you don't inclde all trees), then there is the fact that they had different data to start (e.g., more data perhaps, different species, genes, etc)
15:57 towodo basically there's judgment and evidence selection going on everywhere in this project
15:57 towodo you decide what trees to include, for example (and what taxonomies), and the order
15:57 towodo that's all judgment, yes?
15:58 blackrim well, i don't decide. i facilitate others to decide, but yes, that is judgement
15:59 towodo if we could record all the evidence formally, and turn a crank, that would be more 'objective', but the selection of evidence problem remains
16:00 towodo so what scientific claim do you want to make about the tree?  that it is an informatics artifact reflecting the judgment of many people?
16:00 towodo you will be an author on the paper, so you have to take responsibility for some kind of claim
16:01 blackrim yeah, i don't know if there are any scientific claims we are making about the tree. maybe i am wrong there though
16:03 towodo have you drafted the part of the paper that explains what the tree is and why it is the way it is?  I think any changes to the tree at this point should be made with the explanation of the tree in mind
16:04 blackrim there are models for this in the mammal supertree. for each major group, representative trees were chosen and then a giant supertree (with essentially no conflict) was constructed. ours is in that vein. in the sense that it itself serves as a tool because it reflects something about our knowledge
16:04 blackrim speaking of this paper : http://www.nature.com/nature/journal/v446/n7135/abs/nature05634.html
16:07 blackrim That bit hasn't been written it seems (Karen's name was on it) but I will actually go ahead and flesh it out so that it is clear. Nevertheless, I am unaware, given the goals and nature of the data, of any alternative to the approaches here. I hope alternatives will present themselves as a result of our efforts but we will see
16:08 towodo the question is what is the goal.
16:09 towodo I think it's to record this group's best judgment  (given the amount of time we've invested in creating the tree)
16:10 blackrim right, the goal as I understand is to have a tree that represents the current view (by this group for first release) of relationships of all species (similar to the goal of the mammal and bird super tree approaches).
16:10 blackrim agreed
16:10 towodo That mammals paper says: "All new supertrees were built using an explicit source tree collection
16:10 towodo protocol
16:10 towodo 29
16:10 towodo to minimize both data duplication (for example, where the same data
16:10 towodo setunderliesmorethanonesourcetree)andtheinclusionofsourcetreesoflesser
16:10 towodo quality(forexample,taxonomiesorthosebasedonappealstoauthority)."
16:12 towodo Maybe you could use the paper to make some criticism of supertree methods? Or to talk about just how judgment-infused all phylogenetic efforts are?
16:13 towodo reference 29 is to a book by Bininda-Emonds.  have you seen it?
16:13 kcranstn I tend to lean more to the viewpoint that there will be strange things in the tree that are due to the inputs + methods
16:13 kcranstn if we don't have an input tree that represents the relationship that people want to see, that's just the way it is
16:14 towodo Phylogenetic Supertrees: Combining Information to Reveal the Tree of Life, Kluwer 2004.  There might be claims in there that you could take issue with
16:15 towodo I was telling Stephen I think you should write your methodology section first, then use it to guide tree decisions going forward
16:15 kcranstn then, in the paper, we talk about what would be needed to get to something better (better inputs, better synthesis methods, ability to filter on parts on input trees)
16:17 towodo kcranstn, what you're saying would argue for being able to go back and forth easily in the UI between taxonomy and synthetic tree, better conflict viz, etc.
16:17 kcranstn right. The synthetic tree is only one view into the data
16:19 blackrim sorry, back again and reading through
16:20 blackrim yeah, that is just a general supertree method. of course we have a citation as well for the method, i just mean that olaf picked a very small specific set of trees to include that reflect the (at that point) common view of the mammal phylogeny
16:21 blackrim in other words, it is a very small set of trees
16:23 travis-ci joined #opentreeoflife
16:23 travis-ci [travis-ci] OpenTreeOfLife/api.opentreeoflife.org#401 (nudge-oti-index - b9186a2 : Jim Allman): The build passed.
16:23 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/api.opentreeoflife.org/compare/01f5e6ddd447...b9186a2c4b18
16:23 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/api.opentreeoflife.org/builds/21187507
16:23 travis-ci left #opentreeoflife
16:23 blackrim Anyway, i think we are good, and everything is as expected. We are just going to move forward with the methods and talk about this eukaryote thing on the call i suppose or email
16:23 kcranstn I added it to the agenda
16:23 kcranstn but I do like jonathan's suggestion to spend some time writing up the methods
16:23 towodo blackrim, I think you are being a bit too passive when you say "i facilitate others to decide".  your reputation is on the line since your name will be on the paper. so far you have only given others the ability to nominate and order papers. you could call that a 'protocol'. now we're getting into new territory and you need to put your foot down
16:24 blackrim what do you mean by the last bit, the "new territory" bit
16:25 towodo I mean these two cases that have come up of upper eukaryotes and fungi
16:25 towodo and I'm guessing there may be other pressure to make ad hoc changes to the tree. but maybe not
16:26 kcranstn As always, I am going to resist ad hoc changes
16:26 blackrim yeah, well i have no other choice but to defer for fungi and eukaryotes because i don't know these groups. furthermore, the taxonomy we have for these groups, I presume is bad and I expect changes there.
16:27 kcranstn we do have another choice. We report inputs + methods = results and explain why the results don't always match what we expect
16:28 kcranstn if we don't have an input phylogeny that contains a relationship that we expect to see, that's ok
16:28 kcranstn part of this project was supposed to be highlighting where we need more trees, so if we manually fix the problems, then we lose that ability
16:29 blackrim i agree. this is fine. however, you saw mark's comment on Anolis. imagine the issue if we have no eumetazoa. that would essentially focus the paper on discussing problems with sampling in phylogenies.
16:29 towodo so I suggest write up a post hoc explanation of methods, and going forward only do something outside of that 'method' if you're willing to modify the methods section of the paper
16:29 towodo remembering that the methods section will reflect on your scientific reputation...
16:29 blackrim oh right. i absolutely think manually editing the tree is bad (and actually not possible right now). the only manual editing is taxonomy.
16:30 blackrim there is only manually of any taxonomy
16:31 towodo the eukaryote and fungi problems should be teaching moments. (repeating what Karen said I think.)  maybe highlight them in the paper.
16:33 blackrim that is fine. there is a practical matter that we don't have a usable tool if eumetazoa doesn't exist in our tree because it is generally considered to exist. so we either don't include Laura's tree(s) which is fine and then we can do a separate experiment and section of the paper highlighting this issue, or something else i haven't thought of
16:34 kcranstn usable tool for what?
16:35 blackrim all the stuff people want to use the tree for (ecology, other studies, comparison, etc.).i actually think we do have other evidence for that node, but possibly not for the others (not because it doesn't exist but because it isn't in our treestore or the like)
16:36 kcranstn I think that is ok. If we don't have a tree that contains that node, we don't have a tree. We can ask for one, but we can conjure one out of thin air.
16:36 towodo well… I made the suggestion to extract a bit of SILVA and enter it as a high-priority phylogenetic tree, which it sort of is. I admit that's a kludge and could be considered a 'manual edit'. the fungi case is less clear
16:36 kcranstn I suppose that you could import the greengenes tree
16:36 kcranstn as well
16:37 kcranstn it is a proper phylogeny - i.e. the result of an algorithmic analysis of sequence data
16:37 towodo you mean "can't conjure" not "can conjure", right?
16:37 kcranstn yes
16:38 kcranstn *waving magic wand now*
16:38 blackrim well the fungal one is based on a tree that is citable and based on some sort of manually supertree-like excersize. certainly no manual editing on our part. nevertheless, it is a grey area but one that is easy to explain (I feel).
16:39 blackrim I have not seen the greengenes. is there some reason we aren't using it?
16:39 kcranstn I am working on incorporating it as a taxonomy, but perhaps we should rethink this
16:40 blackrim kcranstn: should we be formatting this for science at this point? I was about to type a bunch and realized that we still had results and discussion instead of the more freeflow science format
16:41 kcranstn we need to send a draft to science first so that they can make the OA decision
16:41 kcranstn I doubt the draft needs to be formatted for submission
16:41 blackrim ok, i will proceed with that in mind
16:42 blackrim i just mean that science doesn't have results and discussion, it is more freeflow than that (proceeds like a story). makes it easier i think
16:42 kcranstn sure, that's fine with me
16:42 kcranstn reformat away!
16:42 towodo methods… write the methods section regardless… you'll feel better
16:42 kcranstn which is supplemental material anyway with science
16:42 kcranstn sigh
16:43 blackrim kcranstn: in regards to greengenes, if it redefines major groups (not gives the relationships but redefines) then it is changing taxonomy. if it is just showing how things are related , then it is a tree
16:43 towodo the people you care about will read it
16:43 blackrim yep, getting methods (and results which is related to that) in there now
16:44 kcranstn http://www.nature.com/ismej/journal/v6/n3/full/ismej2011139a.html
16:45 towodo (I couldn't figure out why they cared about ranks…)
16:45 towodo (you end up with many taxa that only have single children. seems pointless)
16:46 kcranstn they have both a taxonomy and a tree
16:46 towodo I am going to get something to eat before the call. ttyl
16:47 kcranstn ok
16:47 blackrim oh, ok so it is a tree and taxonomy. well then i suppose both is appropriate :-p
16:48 kcranstn the build a tree and then synthesize that with taxonomic info to produce a final taxonomy (sound familiar?)
16:48 blackrim ha, yeah
17:23 blackrim lost sound, did anyone else? do i need to reconnect
18:08 scrollback joined #opentreeoflife
18:16 travis-ci joined #opentreeoflife
18:16 travis-ci [travis-ci] OpenTreeOfLife/api.opentreeoflife.org#402 (nudge-oti-index - d618a64 : Jim Allman): The build passed.
18:16 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/api.opentreeoflife.org/compare/b9186a2c4b18...d618a64ba93b
18:16 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/api.opentreeoflife.org/builds/21196029
18:16 travis-ci left #opentreeoflife
18:19 blackrim kcranstn: still no good solution. i am happy to discuss what I think is the problem more if it is unclear. laura makes a good point but we actually account for that source of conflict
18:30 travis-ci joined #opentreeoflife
18:30 travis-ci [travis-ci] OpenTreeOfLife/api.opentreeoflife.org#403 (nudge-oti-index - db5961c : Jim Allman): The build passed.
18:30 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/api.opentreeoflife.org/compare/d618a64ba93b...db5961c33282
18:30 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/api.opentreeoflife.org/builds/21196956
18:30 travis-ci left #opentreeoflife
18:37 kcranstn blackrim - let's chat tomorrow
19:00 blackrim kcranstn: that is fine. i will keep keeping on with the manuscript. i think we have enough to get most everything down
19:33 travis-ci joined #opentreeoflife
19:33 travis-ci [travis-ci] OpenTreeOfLife/api.opentreeoflife.org#405 (master - 9df49f0 : Jim Allman): The build passed.
19:33 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/api.opentreeoflife.org/compare/f92df31b14a9...9df49f01536d
19:33 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/api.opentreeoflife.org/builds/21201332
19:33 travis-ci left #opentreeoflife
19:48 travis-ci joined #opentreeoflife
19:48 travis-ci [travis-ci] OpenTreeOfLife/phylesystem#176 (master - 7d1505b : Jim Allman): The build passed.
19:48 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/phylesystem/compare/5f7151633040...7d1505bb4825
19:48 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/phylesystem/builds/21201903
19:48 travis-ci left #opentreeoflife
19:49 travis-ci joined #opentreeoflife
19:49 travis-ci [travis-ci] OpenTreeOfLife/phylesystem#177 (master - 9e205f0 : Jim Allman): The build passed.
19:49 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/phylesystem/compare/7d1505bb4825...9e205f03c75d
19:49 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/phylesystem/builds/21201971
19:49 travis-ci left #opentreeoflife
20:50 kcranstn joined #opentreeoflife
21:38 kcranstn joined #opentreeoflife
21:47 dfcnvt joined #opentreeoflife
23:22 towodo yay! neo4j GET methods work!  (still need to add cache control headers)  Boy that was painful.
23:33 jimallman joined #opentreeoflife

| Channels | #opentreeoflife index | Today | | Search | Google Search | Plain-Text | summary