[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: about gsocs2013

Le Wed, Apr 10, 2013 at 10:46:30AM +0200, Nigi Fabio a écrit :
> i' m Fabio, master student @ university of Pisa in computer engineer
> i'd like to have more info about the project on wiki pages for the
> cloud image for bioinformatics with Debian

Dear Nigi,

I was deliberately vague when writing the description of the project for
bioinformatics on the Cloud using Debian, because I want this project to be
bottom-up, where the student particpates in the shaping of the proposal.

I have listed some frequent problems or needs related to the use of Debian
for bioinformatics, which I will detail in a slightly longer form here.

 - When a project starts with a given version of a software, the researchers
   are usually reluctant to upgrade, for good resons as scientific software does
   not have a good record of maintaining stable APIs or outputs.  This means
   that projects end up trapped on a specific computer.  Cloud computing could
   propose a solution to this, maybe in line with the CUT concept, by taking
   advantage of resources such as snapshot.debian.org to deliver reproducible
   systems independantly on the state of the Debian archive.  One could also
   consider how to install multiple versions of a Debian package at the same time.

 - Scientific software also tends to assume that it is run on computers with
   unlimited memory and disk space, which is likely to cause problems on cloud
   systems offering standard units that are often more limited.  This calls for
   parallelisation of the work, specific parametring of the cloud instances
   (which typically have small or virtual temporary directories), scaling
   and cost control, since cloud systems often bill in function of multiple
   parameters such as up time, data transfer and disk activity.

 - Given that the Summer of Code is about programming, what we need to build a
   convincing project that a chance to get a slot is to find a clear target in
   the issues above (or some that I would have forgotten), and propose a solution
   that involves development (without reinventing the wheel) and improves Debian
   in general, while solving important issues about doing bioinformatics in the

I hope this helps,

Have a nice day

Charles Plessy
Debian Med packaging team,
Tsurumi, Kanagawa, Japan

Reply to: