Eric Jain

Curious: Has anyone used BioMoby (especially the service discovery part) for real (TM) applications? - Eric Jain

Same as Eric: I was recently asked "Who use Biomoby" ? - Pierre

There are active communities of users in canada, spain, germany and the UK that I personally know of. Mark Wilkinson could provide you with more specific examples. See also http://pubmed.gov/17237074 and http://pubmed.gov/17496321 - Duncan Hull

Yann: Thanks. I know that many things build upon the arithmetic mean, so I didn't want to completely question its use. I was just curious if there was any other reason (apart from the arithmetic mean being more precise in the absence of outliers in normal distributed data). - Daniel Jurczak

I believe that the mean better represents the distribution of data in large populations/sets of data, while the median is more representative in small data sets. But I am not a statistician. Go ask one. I bet their answer is a lot longer and more philosophical. - Jim Hardy

Daniel: the arithmetic mean is not necessarily more precise, but the statistical test that build on it are, so it is all a question of assumptions - if you assume no outliers use mean and powerful stats, if not use median and non-parametric tests. I guess it all boils down to being consistent, it makes little sense IMHO to mix things. Again, hope this helps! - Yann Abraham

Yep, thanks Yann :) - Daniel Jurczak

Just want to add some stupid ideas: it always comes down to what your dataset looks like. There is no point in calculating statistics when the data has no meaning, is heterogeneous or at least some form where you don't know what to do with it. Personally, I have never liked the median, since it feels so arbitrary and completely independent of the data you have at your hands. - dekay

…oh, and to "answer" your original question: Imagine a series of coin flips. Median or Mean? - dekay

Here's another (purely practical) reason: You want to compute the value at the database level, but (like most relational databases) your database doesn't have a median aggregate function... - Eric Jain

True. But I'd add: 1b. Reuse an existing format (even if it was meant for something entirely different) 4b. Overspecify (to ensure the format can't be used for anything else) 5b. Make it generic (because you're so smart and to ensure it's a pain to use for the intended purpose) 7b. Everyone has dedicated programmers to create tools to hide the complexity of the format from users. - Eric Jain

The format used by the statistical tools performing genetic associations analysises has always been my problem. All those tools use a bunch of variations (HARG!) around a format describing the markers, the individuals and the genotypes. This information is splited across various files (ARGHH !! again) , sometimes lost (alleles are converted into numeric values) and is missing many fields (markers rs##, assembly version etc...) . Tool late: too many tools use this format, it would be hard to change this... - Pierre

I have always preferred a service like Google Docs, at least for the rough draft stages of a paper that is. Cleaner and faster to get started, especially for people who are new to those things. - Daniel Jurczak

http://blindscientist.genedrif... and http://blindscientist.genedrif... - Paulo Nuin

I'll second Egon - though it really shines if you're working in LaTeX. I've never tried to merge two Word docs via subversion. Would it work? - Rajarshi Guha

We are using Google Docs to write a grant, but people who are using it are a little bit skeptical to say the least. It's a slow progress. I never tried to merge Word via subversion, but I guess this is something to check. A quick search showed this: http://nicolas.lehuen.com/inde... - Paulo Nuin

I wish I had posted the whole series I wanted, dunno why I didn't. - Paulo Nuin

multiple editing on a word document can present some issues as it is a binary file, difficult to merge changes via svn. I would recommend LaTex - Frank

If LaTeX is an option then just go for svn. :-) - Daniel Jurczak

@Frank: most of the time, LaTeX is not an option, unfortunately. - Paulo Nuin

Google Docs doesn't seem promising - poor offline version (gears) + inability to tackle anything beyond standard documents (even emf/jpg images do not fit). - Yaroslav Nikolaev

@Rajarshi, Daniel & Frank: what's so special about SVN with LaTeX vs Word? don't have any experience with the former unfortunately...If its only binary vs ascii - one still has ability to track changes/comments from within the Word document, which might be an easier track for an average researcher ;) - Yaroslav Nikolaev

As far as free services are concerned, I've been happy with GoogleDocs for private sharing and Wikispaces for public sharing until the very last formatting step, where we generally use Word. Note that people in my field don't use LaTeX. - Jean-Claude Bradley

@Jean-Claude: what about schemes, images, etc - or you generally don't have a versioning issue on these? - Yaroslav Nikolaev

@Paulo: thanks for the plug, svn-time-lapse looks impressive! - Yaroslav Nikolaev

@Yaroslav, SVN+LaTeX is a nice combination because the latter is plain text. Therefore a simple diff allows you to quickly look at changes etc. I agree that it is not for eveybody - certainly most people in chemistry dept's do not (will not?) use LaTeX. Word's track changes is nice, and indeed I do use that when my collaborators won't use LaTeX - but the lack of branching can be a pain. But in general, I passionately hate Word - anything beyond 5 pages with many figures is simply a pain - Rajarshi Guha

@Yaroslav: My personal caveat with Word+SVN is as you've already guessed the binary vs. ascii thing. Maybe it is just a personal bias. :-) - Daniel Jurczak

More generally, when one writes regularly, the value of focusing on content rather than presentation is a huge boon. And LaTeX documents look sexier than Word documents for the same amount of effort :) - Rajarshi Guha

Its the issue with multiple editors of a binary file in svn - not word vs latex. If two different people check out the same version at 9:00am and work on it. If one person checks back in at 10:00am, the second person trying to check in at 11:00 will get a conflict because svn in not clever enough to resolve or merge conflicts in a binary file. - Frank

LaTeX looks better with LESS effort than Word. Specially now that they abolished menus in Word. - Paulo Nuin

GoogleDocs for the non-Tekkis and LaTeX+VersionControl for the Tekkis, or for the web2.0 freaks a Wiki with RTF export option, this allows post-editing in OpenOffice or Word. One good example is XWiki (Java). - joergkurtwegner

@Frank: this issue is clear! However in non-geeky context it seems easier to resolve it using Word built-in tools, rather than forcing everyone to use LaTeX. - Yaroslav Nikolaev

Any opinions on Git? It sounds more promising in terms of speed & stability (distributed system), however does not seem to have a stable Windows port so far (apart from over-cygwin version)?! - Yaroslav Nikolaev

You could consider using a WYSIWYG LaTeX editor... - Egon Willighagen

Oh, the Git port to Windows is fine, the GUI seems a little bit rough but the command line with SSH included works just fine. The git Eclipse plugin also works fine with most commands, but I had problems pushing things to Github from it. - Paulo Nuin

@Yaroslav For the projects I've been on images and figures don't change enough that it's ever been a problem. On Wikispaces I would just fix the image and replace it. Text is where the massive editing takes place. - Jean-Claude Bradley

One major shortcoming of Google Docs (and most Wikis I've seen) for scientific publications is the lack of support for numbered references - Eric Jain

Eric: that's usually the place were Zotero with its drag and drop comes in. - Daniel Jurczak

Problem with Zotero drag/drop is that it doesn't insert named/numbered citations, just the reference list at the end. The Word/OO plugins are better in that respect, but lag behind the latest Zotero release. I'm sure it wouldn't take too much javascript to get data from Zotero->Google Doc as both citation and reference, if anyone fancies a nice coding project ;-) - Neil Saunders

Neil: Well that's true. To be honest I have never used anything other than LaTeX/BibTeX for "larger" documents, so for me manual adding of references was never a major problem. - Daniel Jurczak

@Eric: exactly the point! that's why file-sharing beats document-sharing in research-paper-writing perspective - one can share a reference library along with the docs.. - Yaroslav Nikolaev

Unfortunately Zotero also suffers from inability to share the library between users/computers..they promise multi-computer sync in Zotero 1.5, and multi-user social sharing in Zotero 2.0...However we're not there yet.. - Yaroslav Nikolaev

Yaroslav: I guess you could use DropBox and put the Zotero library into the shared folder. (??) Have never tried it, so just a guess ? - Daniel Jurczak

The Zotero Sync works pretty well and you can access your database online. - Paulo Nuin

The only thing is, I guess you can't access your papers? Does it also sync up attachments, or at least snapshots? - Chris Lasher

tested Zotero 1.5 Sync (Preview): library metadata and text notes are synced over Zotero servers, while for attachments and snapshots have to use any third-party WebDAV disk (in future developers plan to employ Amazon S3 for this purpose). The problem here is that attachments on WebDAV are stored in own Zotero format (which appears as a collection of .zip and .prop files), so one would have to sync full library with metadata to dig out the appropriate file. - Yaroslav Nikolaev

unconcious bias - although it's about a historical, and not a current problem, a book like Gould's Mismeasure of Man illustrates this beautifully. And, I think that even current scientists need to keep the idea of unconcious bias in their heads at all times :D - Allyson Lister

Outreach. Not enough of it and what there is often not good. - Neil Saunders

true Allyson: an anthropomorphism of science especially, ie understanding how a rat would feel in maze through a human's eyes - sofarsoshawn

I second outreach, ethics and anthropomorphism. good discussion! - Allyson Lister

Definitely outreach from my point of view - so much good work but not clearly communicated to scientists as well as the public! - James Watson

Lack of reporting of negative results- they are usually just as important as positive results. - Shirley Wu

@Shirley -- I hear you, we really need a way for communicating negative results. But how do you really know when a negative result is really negative, and not just a fluke? ;) There's always a better way. You just need to find it. (Or stick with Edison's ways a lightbulb does not work) - dekay

problem to make it productive, profitable and beneficial for society - Alexey

Dekay, point taken - it's true that there are an infinite # of ways something can go wrong, and usually only 1 way it can be right. But I get the impression that a lot of "positive results" are not "typical" - how do we know that the other 100 trials that failed were all flukes? Biologically possible != biologically likely, though I suppose it is also important and cool to see what is possible too. - Shirley Wu

@dekay , @Shirley some sort of clearinghouse for negative results would be useful. You would never know if the negative result was a fluke or not ... but that is taken into account in you assessment (eg, maybe you try the experiment yourself anyway with some tweaks, just in case. If your result is negative too, add it to the clearinghouse - maybe a trend would emerge). A well connected Open Notebook Science 'ecosystem' could play this role nicely. - Andrew Perry

@Andrew Perry I agree, you keep tweaking the experiment until there are negligible "negative results" otherwise you're stuck with a relativist rhetorical argument ie. w/ drug prescriptions there are always negative some side effects, but usually the cure outweighs the disease - sofarsoshawn

Allocating grant money? - Eric Jain

Someone needs to download the NCBI taxonomy database - Neil Saunders

Left a comment at your blog about getting species lists from NCBI. - Neil Saunders

Neil..great source of information. Thanks - Antony Williams

No worries; by the way, taxonomy files at NCBI FTP site are at: ftp://ftp.ncbi.nih.gov/pub/taxonomy/ - Neil Saunders

My argument is that we're in this state because of historical reasons. Much of my argument is here: http://www.davispj.com/bioinfo... There is cursing involved, so don't say I didn't warn you. - Paul Davis

@Paul I'll give it a read. - Chris Lasher

@Paul: Replacing XML with JSON (or YAML) is trivial, but fixes nothing -- unless you think the biggest problem is aesthetics? - Eric Jain

@Eric Not that at all. Its creating an interchange layer that is ubiquitous and easy. XML is the wrong answer for many reasons. Also, I'm a bit biased by working with CouchDB which is why I picked JSON over YAML. - Paul Davis

@Paul I read your article. I hadn't taken a look at JSON until you pointed it out, so thank you. I agree it's a lot less verbose, and aesthetically appealing compared to XML, but--correct me if I'm wrong--it's just an encoding, like XML. It doesn't in itself enforce a standard in the data encoded by it, or the exchange of it, which is the real issue at hand: How do we standardize the data for exchange? How do we provide protocols for the exchange? How do we encourage people to participate in data exchange? - Chris Lasher

@Chris I think the key will be to find some middle ground of RFC's that allow for ad hoc and sub-RFC extensions. Similar to the XMPP specs. Encouraging people to participate is easy. Make it dead simple and show them the benefits. - Paul Davis

Wikispaces+Blogger+GoogleDocs - Jean-Claude Bradley

and as JC points out, plenty of free hosted services that relieve you of most admin and maintenance - Neil Saunders

I would try Google Apps http://www.google.com/apps/int... + blogger - Anders Norgaard

http://modxcms.com/ - Frank

Don't forget Drupal. More work, but powerful. Also evaluate Socialtext and Mindtouch Deki - Deepak

Can't believe I'm beating Ricardo to this: have you considered OWW? - Bill Hooker

I'll also plug OWW, as I'm a user (http://www.openwetware.org/wik...) - but also, they are really amenable to adding features you might like to see (I've had much conversation about searchability). - Heather

I've tried a couple, and second the wiki+blog solution. Unfortunately, unless they have improved it, Google Docs don't integrate well into my favorite blog platform, Wordpress. - Mr. Gunn

If you want something that works out of the box, Confluence may be worth a look - Eric Jain

What value does a blog add that you can't get from a wiki? I find that blog posts have a habit of disappearing into the nether, while wiki pages are productively edited and have a useful outcome. - Donnie Berkholz

@Donnie - for me the value of a blog is that people will subscribe to your RSS feed and you can discuss milestones - RSS feeds on wikis are very messy, except perhaps for keeping on top of seldom used wikis (even there I would prefer email notification) - Jean-Claude Bradley

@Bill I agree OWW is definitely worth considering - Jean-Claude Bradley

I sometimes find it useful to distinguish between blog posts and wiki pages: Blog posts are like wiki pages, except they are timestamped and are not meant to be kept up to date and evolved. Don't want to have e.g. old announcements and reviews clutter your wiki namespace. - Eric Jain

So does anybody use this? Until the 2008 Briefings in Bioinformatics review article by Zhang et al. http://tinyurl.com/5blwsc I wasn't aware such a thing was out there. This concerns me for two reasons: 1) the original paper for DAS was published in 2001 by Dowell et al. http://tinyurl.com/5jvysp and 2) our lab group's focus is in genome annotation. - Chris Lasher

To echo Chris, I'm a big fan of the rationale and goals behind DAS. but if you check out DASregistry.org, the fact that the vast majority of DAS providers are the same groups that wrote the specs doesn't speak to wide adoption. (Does the length 9k+ word spec http://biodas.org/documents/da... have something to do with that?) any comments from DAS users/developers? - Andrew Su

I've tried to use the DAS server at UCSC ( http://genome.ucsc.edu/cgi-bin... ) a few years ago. The result was not enough verbose to use the information. - Pierre

The lofty goals are commendable and something I'm concerned about. I don't believe DAS is the right answer. - Paul Davis

It's not hard to imagine a simpler and more flexible mechanism for finding and retrieving annotations from multiple sources (especially for protein sequences). On the other hand it does seem to be working for some people, and what's the alternative if you don't want to spend a lot of effort creating (and advocating) your own spec? - Eric Jain

I actually believe that DAS is absolutely the right solution for some users. In particular, for example, large genome centers and model organism databases seem to use DAS effectively for data sharing. Someone from (or familiar with) those communities care to chime in? (Todd maybe?) Is DAS just not meant for smaller shops and individual users? - Andrew Su

@Andrew if BioDAS is a spec that requires me to hire a full time person to use it, then something is very, very wrong. - Paul Davis

Paul, I don't necessarily agree. There are specs that take take a full person to implement/use (thing FDA/FAA/OSHA compliance), and rightly so. The target users of those specs can and should devote a person or team to it. But it's always been unclear to me whether DAS is targeted only at big bioinformatics shops (where it does well), or do they actually intend for the average individual bioinformatician to participate (where I think they don't do so well)... - Andrew Su

Andrew, I can't boil my thoughts down to anything more concise than if a distributed annotation system approaches the complexities of governmental regulation its fundamentally flawed. Its connected to: http://roy.gbiv.com/untangled/... - Paul Davis

Implementing the spec does look like a lot of work, but presumably you'd use an existing implementation (e.g. Dazzle or LDAS). There it seems all you need to do (once you managed to set up the software) is figure out how to fit your data into GFF. There may be simpler ways to serve such data, but if there is evidence that there are people who are using DAS-enabled clients and would benefit from having access to your data (or you need a buzzword for a grant application), this investment may make sense... - Eric Jain

Paul, touche... but I chose government as something we'd all recognize. Not sure it was the best analogy. Eric, LDAS seems like exactly what would be needed to bring the complex DAS to a longer tail of developers. It hasn't been updated since 2003 though (http://www.biodas.org/download...), so I wonder how widely it's actually used... - Andrew Su

I was expecting more DAS defenders/educators here. No? - Andrew Su

I've followed DAS quite closely since 2000. I think there's no doubt the spec is top-heavy, especially DAS 2.0. We are developing a genome wiki (http://genome.biowiki.org/) and DAS 2.0's writeback has been mooted as a possibility. Then again there is some investment in DAS (e.g. by Affymetrix) and I expect that when we reached a certain level of functionality, DAS might well look more appealing. - Ian Holmes

Ian, I'm not sure I understand. The demo at http://genome.biowiki.org/test..., does that use DAS to get genome features? Of all the applications, I would think this would be a best-case scenario for DAS use... - Andrew Su

Correct me if I'm wrong, but nowadays I would say "why don't they use RDF instead of DAS ?" - Pierre

@Pierre I'm trying to figure that out, myself. There must be a reason; I must keep reading, keep learning. - Chris Lasher

Can someone walk me through how RDF would to the DAS use case? My understanding is that DAS is great for defining genome features. e.g., Exon 4 of Gene X spans from position A to position B on Chromosome Z. How would one express this in RDF? How would you standardize the vocabulary between all the DAS servers providing this type of data? Just starting my RDF education now... - Andrew Su

Chris, presumably you know the mandatory "restful web services" - Deepak

Take a look at this rails tutorial. Should give you some ideas that should translate to Django? http://darynholmes.wordpress.c... - Deepak

@Deepak I was taking a look at that book, actually. I may pick up a print version of that book, even though VT has Safari access to it. - Chris Lasher

/me bookmarks useful blog post - Neil Saunders

A number of the web services I wrote at IU have REST interfaces - on the backend they do things like retrieve stuff from a DB, calculate molecular descriptors, get 3D structures etc. http://www.chembiogrid.org/pro.... I don't know whether this answers the question, since the input is restricted to a simple identifier - but the descriptor service lets you 'dig' through the various descriptors to get at actual values - Rajarshi Guha

GET requests can still use query string parameters. Something like: http://hostname/by_author?auth..."Author One"&author="Author Two" - Paul Davis

See http://xml.nig.ac.jp/tutorial/... for examples using DDBJ and http://www.myexperiment.org/wo... for a simple workflow that uses it - Duncan Hull

double author query is just a simple GET, e.g. http://www.ncbi.nlm.nih.gov/si...] for Kell + Oliver - Duncan Hull

... http://www.crummy.com/writing/... is a reasonable starting point as is http://www.tbray.org/ongoing/W... - Duncan Hull

If you want to avoid name-value parameter pairs in favour of something more natural and resty, how about http://hostname/authors/smith/... where your app doesn't actually care what order smith & jones come in? - Andrew Clegg

Friendfeed truncated that URL but basically /authors/smith/jones or /authors/jones/smith can be interchangeable - Andrew Clegg

http://www.google.com/search?q... - imabonehead

e.g. http://www.uniprot.org/citatio.... Note that the path is the name of the collection (/citations/), not a verb. The constraints are formulated in a custom (Google-like) query language, and are put inside a single parameter. For simpler cases, per-field parameters may be fine. In any case, what matters is to have a URL that can be GET. Bonus points for supporting machine-readable formats (specified e.g. via a "format" parameter), and offset/limit parameters. - Eric Jain

Don't get too lost in the technicalities or conventions. The HTTP verbs should match the 'intent' of query: for example a PUT translates to a write, a GET nicely translates to a read. With that in mind, GET requests with multiple parameters may retrieve your author list, as may a URL with multiple levels of /author/author. - Matt Wood

Before following the suggestions to integrate all query parameters into the path (especially if their order doesn't matter), read http://googlewebmastercentral..... - Eric Jain

I think there's more of a dynamic-static continuum than that Google post suggests. /authors/smith/jones ought to be *fairly* static even if it's served from a database (assuming unique smith and jones) - Andrew Clegg

Obtaining academic visas don't seem to be a problem (at least from what I've seen). But such places often don't have a budget for flying in candidates for in-person interviews (which most people still prefer). So you'll get invited to a lot more interviews if you "happen to be in town"... - Eric Jain

the visa restrictions make it *very* difficult for companies to consider candidates who don't already have work authorization. First there are an extremely limited number, so you really have to hit the queue in the first day or two the applications are open. Second, there are no guarantees, either of getting the visa or of the applicant following through. (Got burned by one candidate who took another job right after we paid his fees... ugh...) - Andrew Su

Also should mention that at my company, we have hired foreigners at the PhD level, but I can't think of any at a level more junior than that... - Andrew Su

Yes, I have applied to several positions in the US until I found the job I am currently in. I once got a telephone interview, but after that the guy never wrote me back. I don't know if the processing time for an HB1 visa might scare employers (or maybe on in the US thinks I have a good CV). - Paulo Nuin

If the hiring manager is a scientist, then I assure you the last thing s/he wants to think about are the finer points in US immigration law. The system sort of sucks... - Andrew Su

That's what I guessed. But at the same time is quite difficult to understand the "rejection", at least initially, because I was getting interviews in Canada and Europe. - Paulo Nuin

http://bit.ly/2n1PKl... bringing back bad memories. In short, 65K H1B slots, 163K applicants in the first week. Applications open in April for October start. So basically, you have a 40% chance of getting someone who can start in six months... - Andrew Su

more hits in Europe and Canada because caps are looser? Don't really know, just a guess... - Andrew Su

I don't think Canada has a cap for work permits (what they are called here). Usually hiring post-docs is faster than a regular employee, because you need a confirmation letter from HR Canada for the latter. This letter can take up to six-months, and a work permit itself usually takes an extra month. Regarding Europe, I have an EU passport so that wouldn't be a problem, but even with the distance barrier people kept inviting me for an interview. - Paulo Nuin

I do hope it's not a PowerPoint presentation with bullet points ;-) - Eric Jain

I will probably use ppt to call out key themes. Definitely no bullet points. It's hard to appreciate a lot of Tufte's figures in digitized form and the audience is fairly small so a lot of the presentation will be me showing figures from his books and passing them around. - Sutee Dee

Correct me if I'm wrong, but I don't think there's lot of ideas how to apply Tufte's rules to biology. When I compare NYT infographics to finalists in scientific visualization challenges I get impression both images talk to me in completely different languages... - Pawel Szczesny

Biologists tend to be reluctant to generalize/simplify, and therefore often cram a lot of data into their graphs or tables, which makes Tufte happy... Btw does anyone know of a review-type paper that looks at graphics commonly used in biology? - Eric Jain

Eric, I've recently found this: Visual software tools for bioinformatics http://dx.doi.org/10.1016/j.jv... It doesn't necessarily look as what you've described but things like this paper are rarely submitted to journals indexed by PubMed. - Pawel Szczesny

Nyet. - Paulo Nuin

Definitely not - Sally Church

Depends on how much money you have? - Eric Jain

I'd still rather buy (another) Aston Martin or Bugatti - Chris Cotsapas

I don't think I'd even pay $1000 for 500K typing - Chris Cotsapas

I'd like to generalize this question: How much would you pay for sequencing 6 billion base pairs in 1 cell type of yours now relative to your current budget? I'd pay, say $2000. - Attila Csordas

For me, it all boils down to interpretation. Since I don't know what most of it means, why should I pay anything for it? I'd rather pony up the cash (or my genotype, for that matter) to a cohort collection so we can figure out what this stuff means. - Chris Cotsapas

I guess its $399 (for genotyping at least) now and you can download the results. IMO you are part of a big cohort and as an opt in. As long as I maintain ownership, don't mind. $350K though is too hgih - Deepak

I don't want other people's interpretations or choice of markers--I want my whole sequence (every base pair) and I'll interpret it myself. Still, I think it's worth about $20K. - Ruchira S. Datta

I look forward to the day, perhaps some decades hence, when every student in college or even high school gets their personal genome sequence and learns in class how to find information about themselves. - Ruchira S. Datta

There's a bunch of stuff for bioinformatics in Common Lisp. I play with CL when I'm bored - it's very elegant, but I don't enough of it to code fast enough. Also, there isn't a whole lot of library support in cheminfo - I'd love to be able to use it though - Rajarshi Guha

Rajarshi, you should switch you Erlang or Haskell. You're likely to see much better support over time :) - Deepak

Deepak, yes, I ha thought of those two. But a lot of the Haskell related pages I've seen seem to be focused on theoretical CS constructs (though I can see it's getting good library support). As for Erlang, I thought it's claim to fame was concurrency. On the flip side, there are so many dialects of Lisp, it's can be a little painful to write portable code. Right now FP is just a hobby for me :) - Rajarshi Guha

Functional programming (especially with "pure" functional languages) can seem like a bit of a pain. But it makes sense for algorithm-heavy stuff (or so I'm told), and the strict separation of functions with and without "side-effects" is great when you need to write thread-safe code (which I guess we'll have to do a lot more in future). Scala is a nice (non-pure) functional language that conveniently runs in a JVM. Great for learning, but perhaps not stable & mature enough for everyone. - Eric Jain

In Barcelona there is a small group of OCaml enthusiasts http://arxiv.org/pdf/0801.3675 - Anders Norgaard

OCaml is my favorite programming language, but I haven't been using it in my bioinformatics work. There was already a Babel of programming languages in use in the lab when I got here, to which I don't want to add. I do make use of functional programming in Python. - Ruchira S. Datta

For me personally I'm satisfied with the possibilities for functional style programming in Python. - Anders Norgaard

Anders, seconded! I love list comprehensions :) (though proper closures would be nice) - Rajarshi Guha