Camelia, the Perl 6 bug

IRC log for #bioperl, 2009-12-10

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

All times shown according to UTC.

Time Nick Message
00:03 ende joined #bioperl
00:17 ende joined #bioperl
01:45 kyanardag joined #bioperl
02:28 brunov joined #bioperl
03:56 tarbo2 joined #bioperl
05:23 stickyboy joined #bioperl
05:24 stickyboy Who knew you guys had an IRC channel... lucky me.
05:26 ende :)
05:27 ende For some of us, e-mail just aint fast enogh
05:27 ende s/enogh/enough
05:29 stickyboy Freenode is just a cool place to hang out anyways.
05:32 stickyboy I'm trying to come up with an easy way to package Bioperl for my small cluster's compute nodes.
05:33 stickyboy I've resorted to just rsyncing the entire /usr/lib/perl5, /usr/lib64/perl5, and /usr/bin/bp_* stuff to each node after they're kickstarted.
05:33 stickyboy It works fine but I'd prefer an RPM I can roll into my nodes' image.
05:35 stickyboy Also I feel like I'm cheating and it feels dirty.
05:35 stickyboy It's a Rocks cluster (ya know, CentOS 5).
05:35 rbuels it is dirty.
05:35 rbuels lol
05:35 rbuels you installing it from CPAN?
05:35 * rbuels is surprised centos doesn't have a bioperl rpm
05:35 rbuels debian has bioperl packaged
05:36 rbuels but if its a cluster, you might as well just rsync everything to each node
05:37 rbuels alternatively, you can use local::lib to package bioperl into a single dir of its own
05:37 rbuels but then all the scripts would need to add it to their include path
05:39 rbuels rsyncing the entire /usr directories of all your cluster nodes would not actually be out of the question
05:40 rbuels or if you want to get even wilder, version the whole cluster node configuration with git
05:40 rbuels put a .gitignore in / listing /var, /dev, /sys, /proc ...
05:40 rbuels and then cd /; git init; git add .; git commit -m 'initial commit'
05:41 rbuels if you've got like a 5-8G system image, git is actually fast enough to handle that
05:41 rbuels plus for pushing changes to nodes it acts a lot like rsync
05:41 rbuels i haven't actually tried this yet, but i've thougt about it
05:43 stickyboy rbuels: Yeah, and all the nodes are the same hardware...
05:43 rbuels i am sure you are not the first person who has wanted an rpm of bioperl though
05:43 stickyboy rbuels: We use Rocks, so rebuilding nodes is damned easy.  Adding packages too.  The only problem is making a package of Bioperl is a pain in the ass.
05:44 stickyboy Our cluster distribution installs Bioperl, but for it's own version of perl (/opt/rocks/bin/perl), so that causes other problems for the system mod_perl, etc.
05:44 rbuels yes, bioperl's current monolithic packaging makes it a pain in the ass in many respects
05:44 stickyboy Damned if you do, damned if you don't.
05:45 rbuels hmmm
05:45 rbuels well you'll run into the mod_perl issue with any perl modules
05:45 rbuels actually...
05:45 rbuels what is the issue?
05:46 rbuels you *should* be able to just add one or more 'use lib blah blah' into your apache configuration
05:46 stickyboy Hmm.  I should look into that.
05:46 rbuels to pick up perl modules installed elsewhere
05:46 stickyboy I want to avoid complex solutions (like having to remember to modify shebangs for all our scientists' perl scripts/web apps).
05:47 rbuels no, screw that
05:47 stickyboy Or symlinks from one perl to another.
05:47 stickyboy Cuz when I get hit by a beer truck on the highway the poor sysadmin who comes next is going to crap his pants.
05:47 rbuels you can modify where perl looks for perl modules in a number of ways
05:47 stickyboy I'm all ears...
05:47 rbuels environment variable PERL5LIB
05:48 rbuels 'use lib' in any perl that's run
05:48 rbuels -I switch on the perl command line
05:48 rbuels umm...
05:48 stickyboy SetEnv...
05:48 stickyboy Wait a sec, gotta run to another office.
05:48 rbuels compilation options if you are compiling perl yourself...
05:48 rbuels no, don't use SetEnv, use a <Perl> section with use lib
05:49 rbuels <Perl> use lib 'blah', 'blahblah'; </Perl>
05:50 ende joined #bioperl
06:25 stickyboy rbuels: Interesting.
06:59 stickyboy That's a solution for another problem, though... I guess I will just stick to rsync for my cluster nodes.
08:21 bag joined #bioperl
08:40 faceface ptl: sorry, I don't have a consistent 'face' as the teriminology goes.
08:40 faceface I email as dan.bolser@gmail.com
08:41 faceface I was talking about getting an ace file as an alignment object
08:54 ptl faceface: ok, I'll look for it. Thanks. :)
09:07 faceface ptl: yeah... not sure if I wrote anything coherent enough to be actually usefull ;-)
09:07 faceface pretty much my MO
13:09 brandi joined #bioperl
13:31 brunov joined #bioperl
14:14 brandi joined #bioperl
14:16 brandi left #bioperl
16:52 ptl_ joined #bioperl
18:05 zachet joined #bioperl
19:12 kyanardag_ joined #bioperl
19:18 kyanardag_ i have protein sequences and cDNA sequences in hand, i want to extract the CDS sequence from cDNA by using the protein sequence. do you know a script for that?
19:26 deafferret kyanardag_: I'm not sure I understand your question. BioPerl can translate that cDNA in all 6 strands for you, if that's what you're looking for.
19:26 deafferret all 6 frames, that is
19:28 kyanardag_ so, i should translate the cDNA and then match it to known protein sequence and then extract region accordingly?
19:28 kyanardag_ my aim is to extract CDS out of cDNA
19:28 kyanardag_ get rid of UTRs
19:29 deafferret well, I can't tell you what you *should* do.  :)   I can tell you several things about what BioPerl *can* do.
19:30 kyanardag_ is there reverse translate function?
19:30 deafferret um... that lists every possible codon for every single amino acid? no
19:31 deafferret there's billions of reverse translation sequences for even a short protein  :)
19:31 deafferret you could write your own algo to do that
19:32 deafferret bioperl can show you all 6 frames, you can pick the long stretches without stop codons, and assume those might be proteins in real life
19:32 deafferret compare them to a known set, whatever
19:33 kyanardag_ i was trying to avoid writing my own, if there's smt already available.. ok, thanks, i'll do my own in this case..
19:34 deafferret you realize what an exponential problem "all possible reverse translations" is, yes?
19:36 kyanardag_ yes
19:36 deafferret :)
20:24 ende joined #bioperl
20:27 kyanardag_ i'm going with bl2seq with blastx program..
20:44 brunov kyanardag_, about reverse translate, I implemented a routine that does it, it lives in Bio::Tools::SeqPattern, the method is called 'backtranslate'
20:46 brunov kyanardag_, http://search.cpan.org/~cjfields/BioPerl-1​.6.1/Bio/Tools/SeqPattern.pm#backtranslate
20:47 brunov although you won't obviously get a list of sequences, you'll get a single dna sequence that uses degenerate nucleotides, so it will be ambiguous in some parts
20:48 brunov I've been meaning to write something that produces every given sequence from a pattern since then, but haven't actually done it. There's an explanation on how to do it in Higher Order Perl btw, it's just a matter of adapting it to your needs
20:56 deafferret brunov: woof! wow... (LK){2,3}[^GHB][WHYQ]QX.X (YTNAAR){2,3}HBNYRNCARNNNNNNNNN
20:57 deafferret (Backtranslate.t)
20:57 deafferret scary  :)
20:57 brunov :)
20:57 brunov edge cases
20:58 * deafferret is frightened and confused by rampant ambiguity  :)
20:58 brunov it's quite useful to design primers targeting a specific aa sequence (that's what I've used it for)
20:59 brunov but yeah, it gets pretty scary pretty fast
21:03 deafferret (CT).{1,80}(C[GA][GA]CT).(AGGGG){1,200}
21:04 deafferret C_HELP_IM_FRIGHTENED_AND_CONFUSEDCAACT!AGGGG
21:04 * deafferret hides in his safe place
21:04 rbuels lol
21:05 brunov yeah, that's why I wanted a way to take those motifs and get an iterator that would return a 4-letter nucleotide sequence whenever you kick it
21:07 kyanardag_ brunov, thanks.. i'm not trying to find all possible back translations, just back translation of protein with the guide of cDNA sequence..
21:07 kyanardag_ i'll check your code.. thanks..
21:08 brunov oh, then I don't understand what you need, and you shouldn't pay attention at what I said :) sorry
21:08 brunov *attention to
21:08 * brunov goes back to his cave
21:09 deafferret his [cave]{1,200}
21:12 brunov I wish. I barely have (cave){0,1}
21:30 * deafferret rolls the eprimer3 dice, screams about new shoes
23:18 * deafferret goes loco with the cake

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