Camelia, the Perl 6 bug

IRC log for #bioperl, 2010-08-26

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

All times shown according to UTC.

Time Nick Message
04:14 pyrimidine joined #bioperl
04:15 pyrimidine rbuels: just popping in to say, the GFF parser is going to be generic, but has key/value pairs that match the bioperl interface
04:15 rbuels pyrimidine: which gff parser?
04:16 rbuels pyrimidine: it doesn't exist!
04:16 * rbuels chuckles
04:16 pyrimidine Bio::FeatureIO::gff
04:16 rbuels oh.
04:17 * rbuels looks
04:17 pyrimidine so far, I have GFF3, BED, and PTT formats going to the same handler to create generic features
04:17 pyrimidine the data structures are pretty simple, and I need to add URI unencoding, but it's a start
04:18 pyrimidine key thing is, that handler can be subclassed or completely new, just needs to spit back any SeqFeatureI
04:18 * rbuels pokes through his newly clone Bio-FeatureIO
04:19 pyrimidine and (if you don't care about bp objects) call next_dataset
04:19 pyrimidine of course, it needs a shite-load of tests, but it's a start
04:20 rbuels pyrimidine: so if i remember right these modules still have old copies in bioperl-live right?
04:21 pyrimidine yes
04:21 rbuels pyrimidine: which are going to removed after the point release.
04:21 rbuels pyrimidine: right.
04:21 pyrimidine yes
04:21 rbuels pyrimidine: and those are *not* getting the new stuff from this repo merged in?
04:21 pyrimidine right
04:21 rbuels right.
04:21 pyrimidine there is an API change
04:21 pyrimidine which is completely intentional
04:22 pyrimidine the old versions were in prior point releases and were only partially implemented
04:23 pyrimidine rbuels: feel free to hack the bejesus out of them, by the way, if need be
04:23 rbuels pyrimidine: oh yes indeed.
04:23 pyrimidine :) the data structure is still pretty experimental
04:23 rbuels i could probably actually use this gff parser on this particular thingy i'm working on here, since it does not need to be formally deployed
04:24 pyrimidine go for it
04:24 rbuels pyrimidine: any particular test weaknesses related to gff3 you can think of?
04:24 pyrimidine but it would be nice to standardize on something consistent for the data key-val hash store
04:24 pyrimidine URI-related stuff
04:24 rbuels pyrimidine: right.
04:25 pyrimidine older GFF/GTF
04:25 rbuels oh, well we'll let deafferret do that part.
04:25 pyrimidine wfm :)
04:26 * rbuels may do some uri encoding/decoding poking
04:26 rbuels now all i need is a validator that does not blow goats.
04:26 pyrimidine other bits are hooking in Ontology checking
04:26 rbuels i am getting really really tired of the wormbase one.
04:27 rbuels pyrimidine: ok i'm going to take this for a spin now.  i will do my usual project- driven hacking and contribute tests and bugfixes.
04:28 pyrimidine ok.  refine as needed :)
04:28 * pyrimidine goes back to drinking a beer...
04:28 rbuels ttyl
04:28 pyrimidine left #bioperl
05:39 rbuels what conventions have people seen out there for naming the AA translations of a DNA sequence?
05:40 rbuels say you have an mRNA called X, maybe people might call its translations X:tr0 with no frame shift, X:tr1 with 1 base of frame shift, etc?
05:40 rbuels what other things do people tend to do for denoting that?
08:15 kanmani joined #bioperl
08:16 kanmani Hi All...i have a question regarding Bio::Tools::GFF
08:16 kanmani i am not able to get the sequences (dna) corresponding to a feature....
08:17 kanmani while (my $feat = $gffio->next_feature()){my $start = $feat->start;}
08:17 kanmani gets the start....and i can get others like end, frame score etc
08:17 kanmani but not the dna sequence itself
08:17 kanmani any help...
08:28 kai kanmani: is there a sequence in your source file?
08:29 kanmani yes
08:29 kanmani in fasta format at the end of GFFs
08:29 kanmani file
08:30 kai hm, haven't used the GFF parser myself yet
08:31 kanmani can i send you the script...
08:31 kanmani i emailed it to bioperl-l list just now
08:32 kanmani next_feature() method passes next feature....
08:32 kanmani from which i can get start, end etc...
08:34 kai don't see it yet, but that could be the university's graylisting acting up
08:35 kanmani oh! thanks for your efforts
08:38 kai kanmani: I guess once the core devs get up (most of them are in the US) you'll get a reply
08:42 kanmani thank, kai
08:43 bag_ left #bioperl
08:43 kai your script looks reasonablem btw
08:45 kai I see no test for GFF with sequence attached in the source tree, so this might actually be broken
08:47 kanmani ok..
08:47 kanmani how do u do that?
08:52 kai do what?
08:56 kanmani left #bioperl
09:07 bag_ joined #bioperl
09:13 bag_ left #bioperl
11:27 kai oh well, SMILES it is
11:27 kai I can string them together to get a protein
12:04 carandraug joined #bioperl
14:37 dbolser kai: you can type a protin in pymol
14:37 dbolser prolly scriptable too
14:39 kai hmm
14:41 dbolser 07:27 < kai> I can string them together to get a protein
14:41 dbolser what are you doing?
14:43 kai I've got an NRPS cluster that's synthetising a compound. I want to visualize the cluster
14:46 kai something like http://linux1.nii.res.in/~zeeshan/cluster.gif
14:47 kai without the cyclization stuff, that's pretty hard to predict correctly, I know no method to reliably do that anyway
15:05 dbolser cluster?
15:14 kai bunch of NRPS modules
15:14 kai I've been pointed at http://code.google.com/p/rd​kit/wiki/Creating2DPictures, I need to check if I can hack that to do SVG images
15:15 jhamilton left #bioperl
15:16 dbolser bunch of NRPS modules?
15:16 kai non-ribosomal peptide synthetases
16:24 kanmani joined #bioperl
16:43 dbolser oooooooooooh
17:09 rbuels what conventions have people seen out there for naming the AA translations of a DNA sequence
17:09 rbuels say you have an mRNA called X, maybe people might call its translations X:tr0 with no frame shift, X:tr1 with 1 base of frame shift, etc?
17:09 rbuels what other things do people tend to do for denoting that?
17:16 perl_splut I feel sorry for some of the websites that I'm asked to automate using...
17:17 perl_splut I try to play nice with them (putting in sleep statements so I don't just hammer them) and I get chewed out here for my "automation taking too long"
17:19 rbuels perl_splut: maybe they should fix their crappy code so their pages don't take so long to load
17:20 rbuels perl_splut: people do that to SGN from time to time.
17:20 perl_splut the script takes forever to complete because the users here are giving 100k entries to go through
17:20 perl_splut the remote site (which isn't under my control) takes in one sequence at a time
17:21 rbuels perl_splut: can't you just ask them for a dump of their database or somethign?
17:21 perl_splut things like BiMas, Syfpeithi, and Expasy
17:21 perl_splut It isn't data in a database I'm retrieving. it is calculations done to my input sequence
17:22 perl_splut e.g. expasy returns molecular weight, pI, chemical composition, calculation of hydrophobicity, etc...
17:23 perl_splut Bimas and Syfpeithi calculate probability of binding for MHCI or MHCII pathways
17:25 perl_splut I just hate users that have entitlement issues...
17:26 perl_splut I voluntarily do something to help them with their job and instead of a thank you, I get nothing but flack for it
17:26 pyrimidine joined #bioperl
17:28 perl_splut think I'm just going to leave them to do things the hard, slow way from now on
17:29 pyrimidine perl_splut: why not run some of those analyses locally?
17:29 pyrimidine doesn't EMBOSS handle some of these?
17:30 perl_splut well, in the case of bimas and syfpeithi they were getting results from several algorithms and comparing them to find peptides that several sites agreed were good binders, heheh
17:30 perl_splut science by consensus
17:32 pyrimidine tough spot to be in
17:32 perl_splut ya
17:33 pyrimidine we had a postdoc here once who wanted a way to query NCBI in a similar way
17:34 pyrimidine I was thinking it was a one-off thing, turns out he wanted to basically pull down 100K sequences, data, etc
17:34 pyrimidine on a routine basis.  At that point, I left him to his own devices
17:34 pyrimidine :)
17:34 perl_splut heheh
17:36 perl_splut seems the mantra of "Work Smarter, Not Harder" isn't known in the softer sciences like biology...
17:36 pyrimidine depends on how much CS experience the wet-lab person has
17:36 pyrimidine it's better now than it used to be, but it still has a loooong way to go
17:36 perl_splut more like it depends on how much willingness they have for learning something to help them
17:37 pyrimidine most just want a canned black-box service
17:37 perl_splut the people here are expected to be proficient at word, excel and powerpoint as part of their normal job. Most can barely get excel to run, let alone figure out how to get it to help them
17:38 pyrimidine heh
17:38 pyrimidine so, in other words, they are not mentally trained to deal with databases
17:38 perl_splut guess they set the bar pretty low on what qualifies as "proficient"
17:38 perl_splut They won't even pick up a book on Excel to just get better at that
17:38 pyrimidine ick
17:39 perl_splut I've caught a lot of them doing single cut and paste on lines (rinse repeat a few hundred times) rather than click-drag-release from the corner of the cell to copy...
17:39 perl_splut simple, stupid things that are in the first few chapters of any book about Excel...
17:41 pyrimidine the point I recently made to some students here is, if you expect to deal with very large sets of data of any type, learn three things:
17:41 pyrimidine (1) some command line batch stuff (preferably UNIX), (2) a scripting language of some sort (doesn't have to be perl) and (3) how to use a database
17:41 perl_splut I came to this lab with minimal experience in wet-bench work. I can now do anything they can do (though I still have to read the SOP to be sure)...
17:41 pyrimidine much of the wet-bench stuff is cookbook
17:42 perl_splut yep. But I was willing to learn...
17:42 pyrimidine It's a good thing to pick up
17:43 pyrimidine the hard part is having to improvise in the middle of a procedure, or modifying a SOP b/c something doesn't work in a particular step
17:43 perl_splut yep
17:44 perl_splut one of the users came to me with a computer problem this week... The solution, plug the USB cable in to the computer
17:44 perl_splut The error message even said, "can't talk to the device, is the cable plugged in?"
17:45 pyrimidine My tolerance level for that stuff has dramatically declined in the last several years
17:45 pyrimidine I refuse to be the go-to 'computer guy'
17:46 pyrimidine In fact, if I am doing something that is important for publication, I better be on the paper and not just 'we would like to acknowledge so-and-so for help with the computational analysis'
17:46 perl_splut heh
17:46 pyrimidine I did that analysis for you, bub.
17:47 pyrimidine you know, the part of the paper that, if cut out, would cause it to NOT be accepted?
17:47 pyrimidine Can you tell that happened to me once? :)
17:48 perl_splut and then never again
17:49 pyrimidine had another case of that recently here.  a couple of grad students presented a really nice comparative genomics analysis
17:49 pyrimidine found out they didn't run anything, another group of students did
17:49 perl_splut ouch
17:49 pyrimidine and (to top it off) they didn't acknowledge them
17:50 pyrimidine this was raised with their advisor, needless to say.
17:51 pyrimidine Only saving grace was they at least understood (after the fact) what they did was wrong and apologized, but still...
17:51 perl_splut that's pretty bad.
17:51 perl_splut worse than most of the plagarism I've seen in presentations with images.
17:52 perl_splut they grab pictures off the Net and fail to cite the source
17:53 perl_splut just waiting for someone to sue one of the publishers for copyright infringement when papers go to print with their image in it without permission...
17:58 pyrimidine there has been talk hre of having an ethics class for grad students
17:58 pyrimidine *here
17:58 pyrimidine probably a good idea :)
17:59 perl_splut think it is a universally needed class (and repeat refresher)
17:59 perl_splut since some people can't seem to learn the first time...
18:00 pyrimidine well, in Illinois, as a state employee, we have to take a yearly ethics exam.
18:00 perl_splut do all the politicians have to as well?
18:00 pyrimidine well, we happen to have one of the most corrupt state govs
18:01 perl_splut yep. Even the dead vote in Chicago
18:01 pyrimidine (Rod Blago, George Ryan)
18:01 pyrimidine pretty bad when the last two govs have been indicted
18:01 pyrimidine but at least corruption here is bipartisan
18:04 perl_splut corruption everywhere is bipartisan
18:04 perl_splut just look at Washington DC, heheh
18:07 pyrimidine true
18:51 kanmani left #bioperl
18:53 pyrimidine left #bioperl
18:53 pyrimidine joined #bioperl
18:55 pyrimidine_ joined #bioperl
18:58 pyrimidine left #bioperl
18:58 pyrimidine_ is now known as pyrimidine
19:18 pyrimidine left #bioperl
19:18 pyrimidine joined #bioperl
19:19 pyrimidine_ joined #bioperl
19:22 pyrimidine left #bioperl
19:22 pyrimidine_ is now known as pyrimidine
19:34 pyrimidine left #bioperl
19:37 kblin perl_splut: syfpethi can take it
19:37 kblin perl_splut: if not, they should bloody well get a bigger server
19:38 kblin they get the funding for it
20:18 carandraug left #bioperl
22:23 bag_ joined #bioperl

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