Comments on: Apart from Python and Perl in Bioinformatics

By: nuin

nuin — Sat, 27 Oct 2007 15:54:18 +0000

I heard good things about Groovy and even though I am not a Java expert I became interested to learn it. Thanks a lot a for the suggestion and comment.

By: bayesianinitiate

bayesianinitiate — Sat, 27 Oct 2007 05:05:07 +0000

Over time our bioinformatics shop has evolved from an all-Perl shop to using mostly Java for production code. I think it’s more the whole package, the front-to-back suite of Subversion+Maven+Eclipse+JUnit+CruiseControl that has allowed us to efficiently produce robust software for our 24/7 pipeline. Still, there is no getting around all of the QND one-off scripts that we have to write to process some special set of data, and Java is just not good for that. In the past, we mostly dropped over into Perl to do those scripting tasks, which is a jolting change after coding in Java for a few days, but recently we have found something like a state of programming zen by using Groovy. Groovy is essentially java in syntax but with most of the typing noise stripped out. It has very Perl-like regular expression support, and it lets you mix in Java and Java classes freely. It’s slow, but those times when we’re using a scripting language we are out to minimize programmer time more than CPU time. The main point is that since Groovy is so closely related to Java, it lets us do *all* our programming, production pipeline code, cross platform gui code, and one-off scripts in essentially one language.

Groovy is not for you if you don’t do Java programming, but it might be the missing piece of the puzzle for those who already have a Java code base.

By: Bioinformatics » Blog Archives » Apart from Python and Perl in Bioinformatics

Thu, 11 Oct 2007 06:58:29 +0000

[...] Apart from Python and Perl in BioinformaticsA lot of people mentioned Java as a must for Bioinformatics, and I agree with them. But again it wasn’t in the scope of what I intended to write. Even though I don’t like Java (personal taste) I don’t deny its importance in the field, … [...]

By: Tiago Antao

Tiago Antao — Wed, 03 Oct 2007 16:01:03 +0000

One good thing that .NET brought to the world was the idea of the “platform above all”, i.e., people talk mainly about .NET and not about C#, VB.Net, F#, IronPython., etc.

Now I would argue that .NET is not really OS independent (the Mono effort lags and, in practice, Microsoft can render it useless whenever they want). But the idea of “platform above all” is good.

The Java community imported that notion and now there is a framework to plug in other languages to the JVM. This is, in practice, my preferred way of doing things, why?

1. There is a, in practice, really independent and cross-platform virtual machine.
2. The same applies (and this is a most important drawback in .NET) to the libraries. There are public standards for the libraries which are really implemented by many vendors (not to say that Sun’s forthcoming Java 7 will be open source – GPL).
3. Now, which language? Well, I agree, it depends. But, on the JVM (like on .NET) you have options. Big projects with lots of teams? Go to an “overengineered” language like Java itself. Medium sized, small things? JRuby, Rhino(Javascript), Jython, Scala (I will get back to Scala), and many, many others. You can even develop the core of a “big size” application in Java and have a plug in language based on a scripting language (or several ).
4. There is talk about OCaml/F# here. I was an Ocaml user myself (I developed a population genetics simulator in OCaml that I never try to publish). From a conceptual point of view there is a wonderful functional alternative on the JVM, that is Scala. It is not based on Ocaml like F#, but is functional, has most of its syntactic sugar based on Java and is designed from the ground up to work inside VMs like JVM or .NET.
5. OCaml, by the way (I don’t know about F#, as it has nice extensions) lacks something any modern language should have: Introspection abilities.

By: Jon Harrop

Jon Harrop — Thu, 27 Sep 2007 14:42:09 +0000

We’ve sold many copies of OCaml for Scientists to bioinformaticians. I think they really appreciate the combination of Python’s brevity with C’s performance.

You might consider F# as OCaml’s Windows-native cousin. The languages are very similar and share many of the same concepts but F# is much nicer than OCaml on the Windows platform.

By: Lex

Lex — Tue, 25 Sep 2007 14:27:40 +0000

Enjoyed both articles regarding the tools of the bioinformatics trade, and having experience with Java, Python, Perl, and too much R, I’d agree almost completely with your ratings and responses. I have one comment regarding interactive shells. Perls “interactive” shell is actually the builtin debugger (hence the -d flag). It seems to me, after some attempts and frustration, that Perl’s interactive shell is really not as robust as Python’s. For example, in the Perl shell I can’t create multi-line arguments or subroutines, which really limits its utility. Possibly I am doing something wrong? If not, then I think you can put that point back up on Python’s side.

By: Bioinformatics » Blog Archives » Bioinformatics FAQ - Bioinformatics.Org Wiki

Tue, 25 Sep 2007 10:10:51 +0000

[...] Apart from Python and Perl in BioinformaticsI see Java dominating its niche in Bioinformatics, of the large scale, cross-platform GUI application development. At the same time Python and Perl “compete” for their niche of the small to medium applications to quickly deal with some … [...]

By: Robert Pickering

Robert Pickering — Mon, 17 Sep 2007 19:13:50 +0000

As you mentioned F# I used thought I point out one interesting example of its use in this filed. F# has been used by Darren Platt at the DOE Joint Genome Institute to produce a fast and interactive genome viewer, the source code is also short too, only 500 lines.
http://blogs.msdn.com/dsyme/archive/2006/12/06/using-fsharp-to-help-show-we-re-not-neanderthals.aspx