[bip] Blog post on bioinformatics and Python

Nathan Harmston iwanttobeabadger at googlemail.com
Tue Sep 16 02:25:36 PDT 2008


I think the too major factors that keeps Perl firmly rooted in
bioinformatics is:

1) the amount of "legacy" code (imo all Perl code should be termed this).
2) biologists don't like change

Perl became the language of choice due to its strong string handling
capabilities and one a guy during the genome project saying "I know
Perl and its great". Once it got it it stayed in........and I think
we're in a situation analogous to the evolution of the testicular
location from inside of the body to the outside (possibly the greatest
example of how rubbish ID is, apart from the location of the brain).
Perl has been used so much, for so much that it, in order to move back
to somewhere that is infinitely cooler, requires more effort and more
heat and more pain in the short term.

(Most) biologists dont like change..., teach them a language and they
ll use that language for life, teach them a tool and they ll use the
same tool for life. Change the interface and they'll complain. Tell
them about a new paradigm (like SOA) and they will prefer to run
everything as they have been. And to be honest the more time that is
spent learning a new language the less time is spent researching, and
why should they have to.

The only way to make Python better used in the bioinformatics and
systems community is to change the education of biologists and
increase the amount of core library code out there. Next year the MSc
here is going to change to teaching Python and Perl at the same time
(any one with half a brain cell can guess which language people will
like more, and use more). I think python is gaining strength here as
part of its strong maths, modelling, simulation capabilities.

There does seem to be more BioPython traffic out there at the moment
but I've played with it and not liked it, the interfaces, the way it
seems. I did like Pygr but I havent tried using it for a while, but I
dont have the time to even think about asking to contribute to
development ( I m sure I ll have plenty of spare time during my phd ;)
) , and this is a problem I can see for other people as well.

I started to think about a way of getting code together and I did
start to write some of my modules to share with people but didnt get
very far...(I ve got some EBI web services code, but it needs
finishing and testing)  I m a big fan of the well tested code base for
research idea from Titus.........

why dont we start a google project, call it BiP and post code
(unfortunately you won't have any from me atm.....as I'm really hating
R atm, its just like Perl except you count from 1) and post core
modules to it, have a BiP...........force people to provide testing
and docs and post it and move it slowly in an iterative motion
forward. If you give me a few months, I can have a bip.ebi.soapclient
module finished (engineers estimate) and maybe a basic interface to
bioconductor. Anyone interested......non-committal.......no core
direction (yet) just get some community code out there and in use?

And if you use any of the contributed code in research how about
acknowledging the project and the specific groups/people that
contributed the code (I know its not a citation but its the goodwill)
until the point where might snowball into a publication sized project
and cross that bridge when we get there.

Just my few pence, if their worth anything by the end of the year,
damn those investment *ankers!

Nathan

Just another rant!

2008/9/15 Erich Schwarz <emsch at its.caltech.edu>:
> On Mon, 15 Sep 2008, Kevin Teague wrote:
>
>> > the list has been awfully quiet lately.  Here's something to
>> > stir up the rabble:
>>
>> I believe the correct phrasing is to "stir up the pot" and it's
>> this action that "rouses the rabble".
>
>    I myself always thought that the correct expression was
> "harangue the masses" (e.g., "In this 1927 photograph, Vladimir
> Ilyich Lenin harangues the masses").
>
>    At least Lenin didn't have PowerPoint:
>
>    http://www.edwardtufte.com/tufte/graphics/book_pp_cover.gif
>
>
> --Erich
>
>
> _______________________________________________
> biology-in-python mailing list - bip at lists.idyll.org.
>
> See http://bio.scipy.org/ for our Wiki.
>



More information about the biology-in-python mailing list