Camelia, the Perl 6 bug

IRC log for #bioperl, 2011-04-29

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

All times shown according to UTC.

Time Nick Message
00:01 carandraug left #bioperl
00:01 carandraug joined #bioperl
02:16 koviel left #bioperl
03:10 CIA-119 bioperl6: takadonet master * rb27fcda / t/Root.t : added plan (+5 more commits...) - http://bit.ly/jaABoo
05:09 carandraug left #bioperl
05:56 svaksha left #bioperl
05:56 svaksha joined #bioperl
08:12 dbolser rbuels: are you through?
08:35 dbolser heh... it is actually documented! (I guess I never found my way through to the Bio::Location::Split originally
08:36 dbolser http://search.cpan.org/~cjfields/BioPer​l-1.6.900/Bio/Location/Split.pm#strand
08:36 dbolser In get-mode, checks if no sub-location is remote, and if all have the same strand. If so, it returns that shared strand value. Otherwise it returns undef.
08:36 dbolser you live and learn
13:22 pyrimidine yeah, it should probably return 0 in those cases, but I don't like the Split implementation (would rather it be more like a simple DAG than a container with odd behavior
13:30 flu left #bioperl
13:30 drin_ joined #bioperl
13:34 flu joined #bioperl
14:36 zenman_ joined #bioperl
14:56 kai hi folks
15:01 pyrimidine kai: hi
15:01 * pyrimidine is a poet and didn't knowit
15:02 kai oh man, eukaryotes..
15:02 * kai shudders
15:02 pyrimidine just don't call bacteria prokaryotes, there's a man upstairs here who would bring the pain
15:03 kai hm?
15:04 kai I'm mostly annoyed that I have a week to hack in support for genomes containing introns and exons into my software to deal with a reviewer's comment
15:05 pyrimidine kai: that sucks
15:05 kai it seems like all the reviewers we got deal with fungi rather than with bacteria
15:06 kai and trying to argue away all the requested features sounds kind of bad
15:06 pyrimidine well, it all depends on how hard it would be to implement those features
15:07 pyrimidine couldn't indicate the software is limited to bacterial genes?
15:08 rbuels hacking introns and exons into something that's not designed for it sounds hard.
15:08 rbuels i would be pissed too.
15:08 kai pyrimidine: well, part of our frankenstein pipeline supports eukaryotes
15:08 pyrimidine ah
15:09 kai just the parts that I'm responsible for were built for in-house use, and we autoclave everything with fungi in it, not work with it
15:10 pyrimidine still, I would think the caveat could be made that eukaryotic gene models (intron/exon) are not supported at the moment with the current version
15:10 pyrimidine that's essentially how MAKER started out IIRC
15:11 carandraug joined #bioperl
15:11 kai most of the algorithms deal with proteins anyway, so I think it's not too big a change, I just have a PrimarySeqI->translate() call in there that's of course totally off due to introns futzing with the reading frame
15:12 pyrimidine oh, so it's to be something like $sf->spliced_seq->translate
15:13 kai and of course figuring out where in another part of the pipeline _that_ script decides not to serve eukaryotic sequences to me
15:20 CIA-119 bioperl-live: Dan Bolser master * r2e861d6 / t/Coordinate/CoordinateBoundaryTest.t :
15:20 CIA-119 bioperl-live: Adding some tests for Bio::Coordinate::Pair, showing up a bug
15:20 CIA-119 bioperl-live: Signed-off-by: Jonathan "Duke" Leto <jonathan@leto.net> - http://bit.ly/j1pNIV
15:20 CIA-119 bioperl-live: Jonathan "Duke" Leto master * rb48eb8b / t/Coordinate/CoordinateBoundaryTest.t : Merge branch 'topic/bio_coord_pair_tests' - http://bit.ly/iDmQx6
15:24 dukeleto joined #bioperl
15:24 dukeleto blarg
15:26 kai pyrimidine: thanks for the pointer at spliced_seq. this looks safe to use for bacterial sequences as well if I read the code right
15:47 pyrimidine kai: yep, should work
17:06 kblin left #bioperl
17:06 kblin joined #bioperl
17:51 dbolser pyrimidine: DAG?
18:31 pyrimidine dbolser: http://en.wikipedia.org/wi​ki/Directed_acyclic_graph
18:31 pyrimidine but, in truth, even simpler
18:51 dbolser why would splits be DAG?
18:51 dbolser range dag?
19:02 pyrimidine segments can be part of other segments (ex: exons, UTR part of mRNA, which are part of a gene), so there is an implicit parent-child relationship
19:04 pyrimidine s/implicit//
19:06 pyrimidine so, in my mind there isn't any need for a special split class.  A segment can contain child segments; one can add code that checks whether added child segments fit within the bounds of the parent, whether to expand the parent automatically with a child segment, etc.
19:08 pyrimidine there are other semantic issues that would have to be dealt with (strandedness with child segments, what to do with 'remote' locations) but those are addressable in the implementation
19:10 pyrimidine Currently SeqFeature::Generic compounds the problem.  The default SF has-a Bio::Location but can also contain other SeqFeatures (which in turn may have other completely different Bio::Locations).  That's an implementation problem
19:13 pyrimidine IMHO it makes more sense to just have everything be a segment, with a start/end/strand, and have those attributes checked/reset when child segments are added
19:14 pyrimidine but it requires a complete redesign which, with the current code, is a PITA
19:14 pyrimidine b/c it is a change in the expected behavior of Locations, even if the behavior is very possibly wrong
19:15 pyrimidine anyway....
20:03 pyrimidine left #bioperl
20:27 zenman_ left #bioperl

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