DISQUS

Trapped in the USA: Trapped in the USA: Bioinformatics in the Near Concurrent Future

  • kola · 4 months ago
    Well Bosco.. Nice article.. I am a bioinformatician and as with many others am confused with multiple programming paradigms.

    perl, python, ruby are still evolving but have excellent libraries. (ruby least) In comparison java doesn't have many such libraries to crunch bioinfo tasks.

    After seeing all these new paradigms/ approaches, I think I must stick to perl and leave others to air. If we could write clean code in perl, we can do wonders that are not possible in other languages with less code.

    I have recently done 3D graphics in perl (+ OpenGL) and found that perl is god.

    Also, to add, I love php and use it for lot of bioinformatics tasks. I plan to make my own php library for parsing mol files.

    Anyways, appreciate you since you didn't praise scala.(which is the dirtiest possible language)
  • mndoci · 3 months ago
    Bosco, did you ever end up trying out Clojure? Might be an interesting approach to analyzing very large trajectories (driving Hadoop or some other map-reduce engine).
  • boscoh · 3 months ago
    I started a side-project with a machine learning guy in clojure. It was a tough learning curve learning lisp. Unfortunately the project got kind of left on the wayside because splicing databases are just not up to scratch for our problem. It's really nice interfacing with the JVM through clojure. Haven't played around with the concurrency stuff yet. The one downside with clojure is that the error messages are just plain awful.
  • boscoh · 3 months ago
    Did you ever see this DE Shaw paper, which uses mapreduce to analyze very long trajectories? http://portal.acm.org/citation.cfm?id=1413427
  • mndoci · 3 months ago
    Yep. Saw the talk at Supercomputing last year too

    http://mndoci.com/2008/11/24/mapping-and-reduci...