[BioRuby] IRC conference archives
pjotr.public14 at thebird.nl
Tue Dec 21 09:18:37 UTC 2010
Below the first IRC conference in a long time. We are keeping the
#bioruby room open on freenode. Every Thursday 2pm (GMT) we'll try to
This Thursday we can look at developing new plugins. For anyone
interested in some Ruby programming over the holidays.
IRC log started Sun Dec 19 13:51:39 2010
<ChristianZmasek> good "morning"...
<tktym> hi there
<biorelated> Hello everyone!
<pjotrp> the first world wide BioRuby IRC conference :)
<jandot> Hi all.
<jandot> Would it be useful to have very quick introductions from everyone? Less than or equal to 10 words?
<pjotrp> tktym should open first
<tktym> Hmm, I'm Toshiaki Katayama, one of founders of the BioRuby project.
<tktym> Nice to see you.
<pjotrp> Pjotr Prins: 50% BioRuby, 50% other - next year lots of NGS
<jandot> Jan Aerts. Ruby + beginnings of Clojure. Lots of NGS now + visualization.
<ngoto> I'm Naohisa Goto, a core developer, and a release manager of BioRuby
<ChristianZmasek> My work -- comparative and evolutionary functional genomics, software and algorithm development mostly in Ruby and Java (www.cmzmasek.net)
<mishima> I'm Hiro Mishima, using Ruby and Bioruby for human genetics, genomics, and NGS analysis
<mn__> Mitsuteru Nakao: a BioRuby founder, TogoWS, next year lots of SemanticWeb :-)
<biorelated> George Githinji, PhD student, exploring sequence diversity and immunoinformatics next year
<tktym> hello, we've just finished self introductions from each other
<yannickwurm> arg -sorry I was stuck with something. Is there a log already?
<tktym> I'm not sure. Pjotr?
<pjotrp> I am logging
<yannickwurm> my miniintro: I'm mostly a bioruby user. I'm an evolutionary biologist trying to understand the genetic basis of social life in ant colonies.... after getting fed up with perl's weirdness I decided to use bioruby for most of my scripting
<tktym> ant colonies! sounds interesting
<tktym> OK, how do we start discussion?
<jandot> Is Raoul coming on the call? Would be useful for the plugin stuff.
<tktym> Probably, he is in Spain for his vacation.
<tktym> As for the plug-in system, we (ngoto, mn and me) tried to setup bio-geek and started to make the first plugin, last week.
<jandot> How did that go?
<jandot> And what plugin was that?
<tktym> Installation procedure was fine.
<tktym> I thought we should use a name "biogem" instead of the "biogeek".
<tktym> ruby <-> gem ; bioruby <-> biogem
<tktym> for symmetry
<jandot> I'd vote for that as well.
<tktym> and bioruby plugins are actually RubyGems.
<tktym> OK, I'll talk with Raoul on this.
<tktym> Then, biogem (biogeek) command created a template plugin directory structure.
<tktym> It is working very well, but we may need more templates with a lot of comments instead of empty files.
<pjotrp> the plugin system is looking good
<tktym> from a simple version (ruby only, 1 file) to a complex version (many files and may contain C code)
<tktym> The version we tried is the former, simple one, which was kind of Bio::ColorScheme extension.
<tktym> more info: http://bioruby.open-bio.org/wiki/Plugins
<tktym> but I'm not sure these descriptions are the latest version ;-)
<ngoto> It seems not up-to-date
<tktym> OK, I'll contribute my installation log there.
<tktym> Anyway, my overall feeling about the plugin system was good and it seems a way to go.
<tktym> Additionally, I'd like to discuss about the inclusion of workflows (with data) as plugins.
<ChristianZmasek> workflows sounds interesting...
<pjotrp> how do you see that?
<tktym> there are a lot of methods described in journal articles, and probably some of them can be coded in (Bio)Ruby
<tktym> if the procedure can be distributed as a plugin,
<tktym> (with example data)
<tktym> user can try the workflow with ease and with his own data as well
<jandot> That'd be really cool. Would fit in nicely with the new journal Open Research Computation. (disclaimer: am an editor)
<ChristianZmasek> That'd be really cood -- agreed!
<pjotrp> integrating git for history?
<jandot> One of the issues will be that many workflows rely an very specific site-specific setups.
<tktym> I think this kind of plugins are something like R's package, which is one of the strength of R and BioConductor.
<pjotrp> yes, which also allows people to submit to journalx
<tktym> Jan, yes, but we can start with generic ones.
<pjotrp> I think what is required now is a few examples
<jandot> NGS-related would be nice :-) But that is at the moment not the strongest field in bioruby... Something phylo?
<tktym> yes, NGS with stats and/or visualization would be a candidate. GFF3 too.
<pjotrp> what NGS formats are you using?
<ngoto> products of Google Summer of Code 2010: NeXML, gene duplocation algorithm
<jandot> fastq -> sam/bam -> pileup or VCF. @pjotr: biolib?
<pjotrp> biolib has sam/bam (samtools)
<pjotrp> biojava too
<pjotrp> biolib version is not complete
<tktym> What was in my mind : analysis methods (>200?) implemented in G-language http://www.g-language.org/wiki/documents
<jandot> picard = java library on top of samtools
<jandot> All: I have to rush off, unfortunately. @pjotr: will you send us the logs?
<pjotrp> yes - both biolib and picard are samtools
<pjotrp> I have made screencasts for JRuby
<pjotrp> but I have trouble getting the sound track right
<pjotrp> silent movie now
<pjotrp> anyway, coming soon
<tktym> can we watch the silent version?
<pjotrp> maybe not that good, it needs info
<tktym> I could not spare time to watch the movie, Raoul sent us a link before. http://vimeo.com/14435288
<tktym> Hmm, "Rails on the JVM with JRuby"
<pjotrp> it is mostly about JRuby and tools
<tktym> worth watching? it seems the length of the movie > 1h!
<pjotrp> it gives an idea of JRuby and the JVM tools
<pjotrp> I think the JVM is a good way forward for BioRuby
<tktym> for BioRuby developers or users?
<pjotrp> developers benefit from existing libraries and tools
<pjotrp> picard a point in case
<pjotrp> users benefit from packaged jars
<pjotrp> you can package everything together, including
<ngoto> Is a ruby program packaged by using jar?
<pjotrp> jruby, scala, biojava, bioscala and gems (!)
<tktym> I'm still not sure because I suppose that most of BioRuby users are not that geeks ;)
<pjotrp> yes - that you can see in the video, all in one
<pjotrp> tktym: it is easy for the users, even one-click installs on most platforms
<tktym> OK, potentially it sounds very interesting.
<pjotrp> only thing that causes problems are gems with C libs
<pjotrp> but, I think we can avoid that for most use cases
<pjotrp> also the JVM is very fast these days
<pjotrp> there is no reason to program in C any more, except a few border line cases
<tktym> so, what kind of help do you need?
<pjotrp> I am definately heading this way for big data
<pjotrp> no help, per se
<ChristianZmasek> I'm still not sure because I suppose that most of BioRuby users are not that geeks -- I would very about too...
<pjotrp> I am interested in finding people
<pjotrp> who are interested
<pjotrp> ChristianZmasek: what are you saying?
<tktym> Anyway, we can try both with plugins. Some plugins for high-end users can depend on JVM to BioScala but we can still use BioRuby as it is now.
<pjotrp> we can take it case by case
<pjotrp> ruby 1.9 is pretty fast
<pjotrp> jruby will win with parallel programming
<pjotrp> that is my main focus
<ChristianZmasek> pjotr: to me it sounds as if these recent proposals (also on the mailinglist) are moving away from what many like about Ruby: simplicity and elegance.
<pjotrp> ChristianZmasek: I think bioruby on its own is a lost cause
<pjotrp> we just don't have the momentum
<pjotrp> biopython is going to be successful
<pjotrp> bioperl may live for some time
<pjotrp> biojava is doing really well
<pjotrp> so where are we?
<ChristianZmasek> wouldn't worry that by spreading out the few developers over more "projects" would make it even worse?
<pjotrp> ultimately bioruby is interpreter agnostic
<ChristianZmasek> don't you think it is possible to speed up many/some things from within bioruby (more effiecient coding) than relying on numerous "outside" tools?
<pjotrp> a lot needs to happen with bioruby, to make it suitable for big data
<pjotrp> not only that
<ChristianZmasek> I remember somebody writing about some module for numeric methonds in ruby -- that sound interesting...
<pjotrp> ruby/python/perl are not fast enough for churning 1TB
<pjotrp> (it is not only an IO problem)
<pjotrp> rbnum would be useful, though we have quite a bit of that already
<ChristianZmasek> ruby/python/perl are not fast enough for 1TB -- agreed! Wouldn't even think about using scripting languages for that amount of data...
<ngoto> how about Java/JVM?
<pjotrp> Java/Scala are strongly typed
<pjotrp> that means they can be highly optimized for speed
<pjotrp> The JVM used to be bad with memory
<pjotrp> but that has mostly gone now
<pjotrp> but I predict that people will go by use cases
<tktym> I feel the amount of data itself is not a problem of coding. We can use DBs and indexing systems optimized for that. If so, all we need inside bioruby is binding for those external data store?
<pjotrp> if I write a fast GFF3 Scala lib, that you can use from bioruby
<pjotrp> are you going to reimplement that?
<pjotrp> tktym: that requires DBs etc.
<pjotrp> do you really think a JVM dependency is harder for users?
<tktym> I always use DBs or TokyoCabinet or other indexing systems for speed.
<pjotrp> Most machines have a JVM now.
<pjotrp> 1. is it user friendly?
<tktym> I wrote: Difficulty of depending JVM is something equivalent to have external DBs for ordinary users in this context. It depends, and it would be a requirement for some cases.
<pjotrp> sorry, pressed the wrong key ;)
<tktym> I think JVM is one way to go, but I also suppose plugins and workflow applications will attract people if we can focus our efforts to making really useful ones for daily research.
<tktym> Anyway, we can continue both approach.
<pjotrp> i disagree that JVM makes it harder. It makes it even easier that what we have now
<pjotrp> you'll see.
<tktym> OK, I'm looking forward your movie :)
<pjotrp> well, they are dev focussed
<pjotrp> I'll need to make another one (#4)
<tktym> We spent some time on the topic of JVM, I'd like to take from others.
<tktym> Just for statistics and to plan the next IRC meeting, can I ask where are you connecting from?
<tktym> tktym, kozo2, ngoto, mn are from Japan. mishima as well?
<tktym> I remember someone from Australia wanted to join this meeting if the time is sane. :)
<tktym> (As I joined conferences in Japan for past two weeks, I could not reply the thread on the mailing list but I just finished to read them yesterday.)
<tktym> We have 12 members (+2 raoul and jan) on this channel.
<yannickwurm> i'm in jakarta indonesia
<yannickwurm> 30 degrees celcius actually :)
<pjotrp> I am near Amsterdam, Netherlands.
<pjotrp> welcome :)
<francesco_strozz> I am late, sorry : (
<tktym> Today, we talked about BioRuby plugin system (current status and future) and BioRuby with JVM for high performance computing.
<tktym> Probably, Pjotr will put a log on the web?
<ChristianZmasek> I am in California
<pjotrp> tktym: yes
<tktym> Chris, what time in your time zone?
<pjotrp> ChristianZmasek: early bird
<ChristianZmasek> I can be productive today!
<pjotrp> yannickwurm: what are your bioruby needs?
<yannickwurm> pjotr: work with ants is a bit different because you need to do everything yourself (eg: there is no established GO annotation)
<pjotrp> are you sequencing
<francesco_strozz> yannickwurm: that's the same in livestock animals for me
<yannickwurm> so I've been parsing a lot of stuff (blasts...) and creating small pipelines to stick different tools together, and sometimes run things parallel (using only ruby threads for now... maybe the rqueue in the future)
<yannickwurm> we sequenced and assembled our genome (will be published in PNAS in january :))
<pjotrp> we all seem to be doing pipelines
<yannickwurm> so have been dealing more and more with Illumina sequence (fastq), 4 sequence (SFF), their assembly, their mapping (for RNAseq stuff)
<ChristianZmasek> I am in "Pacific Time Zone"
<pjotrp> yannickwurm: from Indonesia?
<yannickwurm> I'm usually at the borderline between "shell script" and "pipeline" (most are one-off)
<yannickwurm> I'm a postdoc in switzerland
<yannickwurm> but in september my partner and I moved to indonesia
<pjotrp> ah, holiday
<pjotrp> too quick ;)
<yannickwurm> no - well be here for a year or 2
<yannickwurm> I'm still using the Swiss cluster though (www.vital-it.ch) :)
<ChristianZmasek> yannickwurm: I am Swiss as well
<yannickwurm> gruetzi :)
<pjotrp> I have lived in Bangladesh, Italy and UK
<francesco_strozz> piotr: n Italy also? Where?
<francesco_strozz> (I'm italian)
<pjotrp> My wife was working with UN FAO
<pjotrp> May happen again
<francesco_strozz> I see
<pjotrp> Where are you in Italy?
<pjotrp> Near Raoul
<francesco_strozz> near Raoul
<pjotrp> Should meet up at ISMB Vienna
<pjotrp> I am planning to take the motor bike - Guzzi
<francesco_strozz> in july? Is with ECCB2011 if I remember
<pjotrp> yannickwurm: where in Indonesia
<pjotrp> it is kinda big
<yannickwurm> jakarta crazy
<pjotrp> yannickwurm: I have an Indonesian friend there
<yannickwurm> biorubyist? :) I'll hire him :p
<pjotrp> heh, no programmer
<pjotrp> we'll take this offline :)
<pjotrp> It is getting late for some of us (Japan)
<tktym> Yep. 1AM now
<pjotrp> Maybe we should do this more often?
<tktym> Shall we regularly continue this kind of meeting on IRC?
<yannickwurm> yes, I have got to get going too
<mjy> yannickwurm: quick question
<mjy> do you know about the Hymenoptera Anatomy Ontology? (looking at your solenopsis now)
<pjotrp> who would not join in next time?
<yannickwurm> mjy: heard about it, but I'm not a taxonomist so haven't pushed much to find out more
<pjotrp> the thing is
<mjy> yannickwurm: we should chat sometime, it's much than taxonomy
<pjotrp> the 'core' team does meet up occasionaly online
<mjy> erm much *more*
<pjotrp> we feel it is useful to broaden the meetings
<pjotrp> but the timezones are difficult
<yannickwurm> pjotr: with fixed agenda? or as 'bioruby outreach'?
<pjotrp> I think of it as a meeting of minds
<pjotrp> so, it is more outreach
<francesco_strozz> will you send a copy of the log through the mailing list ?
<pjotrp> already it is useful
<pjotrp> francesco_strozz: sure
<ChristianZmasek> already it is useful -- yes
<pjotrp> one thing we could do is weeklies
<pjotrp> keep IRC going anyway, but meet up, say one day a week
<pjotrp> tktym: what is a good time for you, to be logged in?
<pjotrp> ngoto: same
<pjotrp> probably not Friday night :)
<pjotrp> maybe Thursday?
<tktym> We are basically online and we can setup a IRC bot to always keep this channel open and to make the log on web (that we had been doing before the Skype era), so all we need to set "core time" - Thursday maybe fine.
<pjotrp> 11pm JST, 3pm CET, 6am PST every Thursday, does not have to be active conversation, just representation of BioRuby
<pjotrp> Time to discuss developments, frustrations, and encourage new developments?
<pjotrp> ChristianZmasek: OK?
<yannickwurm> and "I'm a n00b user why doesnt XXX work?" ?
<tktym> anything :)
<francesco_strozz> that's sounds good :-)
<ChristianZmasek> OK -- Th morning (for me)!
<yannickwurm> that would be good - and important to advertise on the website
<tktym> the key point is, the conversation log will be made public on the web
<yannickwurm> because the first "im going to try something" is when people either stay or leave
<pjotrp> we need to recruit new contributors, and find ways of leveraging existing code and projects
<tktym> I hope everyone on this channel will help answering those newbie questions :)
<pjotrp> like rtfm
<pjotrp> rtfc is more likely with BioRuby ;)
<pjotrp> Students may have been exposed to RoR
<pjotrp> so, yes.
<yannickwurm> Hehe - anyway I really gotta run. I would like to just say that I am thankful for all the efforts that have been and continue to be put into bioruby. While many aspects clearly have potential for improvement, I differ with Pjotr's view that bioruby may be a lost cause. My vision is limited, but I have the feeling that bioruby has a solid base and hope that positive energy will lead to more contributions & improvement in the code, as well as m
<yannickwurm> thanks & good night!
<yannickwurm> ok i'm going to control-c
<francesco_strozz> piotr2: sorry if this was already discussed but, why don't put (maybe on the website), the "open issues", so libraries that need improvements or new libraries to be written so people can see what is needed and maybe propose to help or write parts of bioruby ?
<ChristianZmasek> and important to advertise on the website -- agreed
<tktym> francesco: https://github.com/bioruby/bioruby/issues
<pjotrp> it would be good, yes, but experience tells me it is hard for people to (start) contributing
<tktym> existing issues can be found on the above page,
<francesco_strozz> tktym: didn't knowâ¦.thanks :-)
<tktym> assigning "new libraries to be written" for others is a hard task
<pjotrp> There is more to it: I have quite a few branches of bioruby myself, which never make it into mainline
<tktym> we need to mentor them, we need to know their ability and motivation etc.
<pjotrp> that is one reason to have plugins
<tktym> some such chances were Google SoC
<pjotrp> so we can have 'experimental' code with bioruby
<francesco_strozz> tktym: yes I know, can be a hard task
<pjotrp> I think development is driven by the 'itch' that needs scratching
<tktym> pj: yes, I believe plugin system will casually invite more people to contribute their code to bioruby
<pjotrp> it is easy to fork bioruby or plugins
<pjotrp> but it does not necessarily translate into active and coordinated contribution
<ChristianZmasek> I still need to write to promised phylogenetics related tutorials... :x
<mjy> ChristianZmasek: I would like that :)
<pjotrp> but let's continue with these IRC conferences, at least to encourage contribution
<pjotrp> we are a social species...
<tktym> like ant... :)
<mjy> Is there a "use-case" list for desired bioruby features?
<ChristianZmasek> I still need to do some cleanup on the Google Summer of Code - "duplications" project and then it needs to be merged... :x
<tktym> mjy: probably, no
<pjotrp> merging is the hard bit
<pjotrp> (mad laughter)
<tktym> mjy: http://bioruby.open-bio.org/wiki/Plugins#Candidate_Whish_List
<pjotrp> let me ask, who would like to contribute to BioRuby itself here?
<mjy> I would at some point- don't know if I have the skills yet :)
<francesco_strozz> me (if I meet the required skills, otherwise I will improve skills and then come back to say "me")
<pjotrp> heh, that is quite a few who have not commited yet
<pjotrp> I don't think skills should be that critical
<pjotrp> we all learn as we try
<pjotrp> plugins can help - even to replace core code
<pjotrp> maybe each of us should start thinking in terms of creating a plugin
<mjy> pjotrp: am I correct in surmising that plugins are just code with a Bioruby namespace?
<tktym> mjy: these items in the plugin wish list are mainly libraries for file formats and external programs, but I would like to propose researchers to provide "workflow" plugins to reproduce the scientific procedure.
<mjy> just read the "What is a plugin"
<pjotrp> it can be really loosely connected, or tight bioruby
<pjotrp> I wish to encourage you to come up with an idea for a plugin
<pjotrp> we will support you
<mjy> pjotrp: ok- so I have several gems that I used jeweler for- it would be nice there was a "jeweler" for bioruby plugins if you konw what I mean
<francesco_strozz> tktym: I think workflows could be a great plusâ¦.if you think about all the pipelines we need to write to run, for example, sequencing data
<pjotrp> That will come.
<tktym> francesco: yes, that will be a one of the goals :)
<pjotrp> tktym: is the idea gridengine or TORQUE?
<pjotrp> to cluster compute stuff?
<ChristianZmasek> tktym: I think workflows could be a great plusâ¦ yes!
<pjotrp> tktym: a workflow system design could also be a paper
<pjotrp> anyway, let's continue Thursday?
<tktym> Currently, real usages of the bioruby library are buried in the test codes, but we can bring it to front as a plugin.
<tktym> I'm fine with Thursday
<tktym> Besides, I'll setup "nadoka" http://www.atdot.net/nadoka/ as before.
<pjotrp> bye! And thanks!
<tktym> which is a IRC bot written in Ruby :)
<tktym> by the core developer of Ruby-1.9, Sasada-san !!
<pjotrp> I'll get the audio on the screen casts done
<francesco_strozz> bye, see on the next irc call
<tktym> bye for now
<ChristianZmasek> bye and thanks!
<tktym> hello bioruby
<tktym> hello again (testing)
<tktym> OK, I think http://bioruby.org/irc/ is ready for you :)
<tktym> Updated http://bioruby.open-bio.org/wiki/BioRuby_IRC_conference
<tktym> Updated http://bioruby.open-bio.org/wiki/Plugins#Develop_a_plugin
<tktym> and added http://bioruby.open-bio.org/wiki/BiogemInstallation
IRC log ended Sun Dec 19 22:50:34 2010
More information about the BioRuby