Re: Do we need Debian(-Med) machines?
Le Mon, Aug 06, 2007 at 11:56:15PM +0200, Steffen Moeller a écrit :
> Hello,
>
> in Vienna I had the opportunity to present the Debian-Med page and Debian in
> toto to representatives of IBM, SGI and HP. Guess what - they all liked it a
> lot, none knew about it. With some caveats that they might like about
> everything at such occasions and as a non-native English speaker I might have
> missed something, I truly feel that we could get (access to) hardware for
> Debian fairly easily, particularly here to mention are PowerPC-compatible
> Cell machines.
>
> What and how much they would give us would depend on how we present ourselves
> to them, which again depende on what we would need them for, of course. Do we
> need such machines in the first place and if so, what would be needed?
Hi all,
One thing that may be ressource-hungry in the future would be the
generation if index files for sequence databases. With the new dbx*
instructions of EMBOSS it takes ages, so definitely we could add value
by distributing indexes in .deb pacakges and let the users or the
package itself download the sequence databases since they can be truly
enormous. I do not know if doing that kind of calculations on debian.org
machines would be welcome, so this is one example of use we could make
of their machines.
To better advertise ourselves and the free software in our fields we
could also set up a ressource website using Debian-Med. This is of
course more tricky as "web ressource" means "millions of script-kiddies
eager to take control of the machine". However, it could be more
valuable for the hardware sponsor as it would give customers an idea of
the kind of machine they should buy to set up a similar ressource. In
the genomics field one obvious thing to set up would be a genome
browser, but there are already many excellent ones. On a more modest
scale, we will have soon in Debian an enhanced version of M-Coffee
kindly prepared by Cédric Notredame. M-Coffee is a meta-aligner which
uses the output of may programs, but some of them are non-free. In his
latest version, there is the possibility to replace clustalw and by
kalign, which is packaged in main.
In summary, here are the suggestions I made :
- Power-hungry generation of .debs containing indexes for sequence
databases;
- Web ressource using `DM-Coffee', a modified version of M-Coffee for
Debian, which uses only programs from main.
Have a nice day,
--
Charles Plessy
http://charles.plessy.org
Wako, Saitama, Japan
Reply to: