Dear all,
over the last weekend in January the Debian Med team hosted the first
Sprint on Bioinformatics [1] in Travemünde at the Baltic Sea in
Germany. It was an intense time of up to 6 Debian developers (tille,
moeller, banck, manuel, cts, schuldei), another two with bioinformatics
packages already in the distribution, a representative from Ubuntu derived
Bio-Linux [2], another from Knoppix derived Goebix [3] and then all the
bioinformatics researchers, teachers and sysadmins that totaled to 25
individuals from 6 countries.
The meeting had a couple of very tangible outcomes
* The packaging of Ensembl [4] (joint project by the EMBL - EBI
and the Wellcome Trust Sanger Institute to produces and maintains
automatic annotation on selected genomes) was continued and
uploaded to the new queue for experimental. It is prepared by and
maintained with all insights from the company Eagle Genomics [5]
in Cambridgeshire. They benefit from it themselves when setting up
their cloud environments with Debian and so shall everyone else.
* New packages
- Bowtie
- TMalign
- rq
- vienna-rna
- mothur
* Revisions in collaboration with upstream
- T-Coffee, stimulating the packaging of many additional
alignment tools
* We got cran2deb [6,7] installed on
master.dermacloud.uni-luebeck.de/cran2deb/rep
and ran through the 2565 most-easy-to-pack packages from CRAN [8]
and another 898 from BioConductor [9] over the last two weeks.
* Taverna [10] could already talk to Debian's command line before,
but now it can also understand the description of tools in EMBOSS
[11] in an automated manner. This involved our community and
the representatives from both upstreams alike.
But we have met for the soft sides more, where we have to report on
* Training on packaging software for Debian on 6 individuals. But as
usual, on some higher level everyone learned something from everyone
else. Key signing. Hey, quite some of us where seeing each other
for the first time. This helps.
* We learned about how Debian Med is perceived. There should be less
geeky web pages or those that we have be made more obvious. We are
doing many things right, but we should advertise them more. We were
not completely confident that we have a critical mass of developers
already. But we can definitely say that as community Debian Med
has long exceeded any regular academic funding scheme.
* Discussion of differences and complementary ideas between getData
[12] (low tech and close to Debian) and biomaj [13] (very user
friendly but no direct interaction with Debian packages) for the
automated retrieval of biological data and its post processing.
* EMBOSS and BioLib [14] found the right level of abstraction to get
linked up.
* Bio-Linux and Debian Med have found ways to collaborate (and have
done so with T-Coffee and TMalign already). This will be particularly
interesting since Bio-Linux is allowing itself not to completely
adhere to the DFSG when this helps packages to be used by their
community.
We have decided to mutually learn from our experiences here and
communicate the problems or successes with repackaged software. This
mostly addresses e.g. Java programs in the bioinformatics community
which frequently depend on a very particular version of a library,
and even more often the exact version of a particular Jar is not
even known. Bio-Linux can in those cases ship a binary package. We
from Debian Med cannot upload binary only packages to main and we
do not want to use non-free in this case, even though we could.
It would be in our interest to explain Bio-Linux as an extension
from what Debian Med is offering. We will be formulating something
together in this respect over the next days.
* We are still discussing the effect that non-redistributable
software (you can download binaries only from the developers but
are not allowed to give that to someone else) but with inspect-
and buildable sources (you are asked to send patches in) should
have on the Debian Med repository. The situation is a bit like
the opposite to having only binary packages in Bio-Linux.
In the subversion and git repositories of Debian Med we can
exchange the code that is required to prepare Debian packages
for such software. The source code or the binaries themselves
are not shared. Some such software is rather delicate to build
and the formalities of the build process to be exchange publicly
through us should play well into the hands of the respective
communities.
The NERC Environmental Bioinformatics Centre in Oxford [15] that is behind
Bio-Linux has direcly paid for the travel of multiple participants for
~1000 pounds. This was very generous and we very much thank Tim and the
Bio-Linux team for this. We should also consider that as an indication
of the special commitment that the Bio-Linux group is preparing for. The
employers of the other participants have funded the others' respective
travel. Many thanks also go to them. The DDs either came from the
region or shall have their costs compensated by Debian funds, which then
shall remain well in the budget forseen for the 5-7 people originally
planned for.
Many thanks to all the participants for a tantalising time together,
in the name of all of us
Steffen and Andreas
[1] http://wiki.debian.org/DebianMed/Meeting/Luebeck2011
[2] http://nebc.nerc.ac.uk/tools/bio-linux/
[3] http://www.bioinf.med.uni-goettingen.de/teaching/goebixdvd
[4] http://www.ensembl.org
[5] http://www.eaglegenomics.com
[6] http://debian.cran.r-project.org
[7] http://master.dermacloud.uni-luebeck.de/cran2deb
[8] http://cran.r-project.org
[9] http://www.bioconductor.org
[10] http://www.taverna.org.uk
[11] http://emboss.sourceforge.net
[12] http://wiki.debian.org/getData
[13] http://genome.jouy.inra.fr/biomaj
[14] http://biolib.open-bio.org/wiki/Main_Page
[15] http://nebc.nerc.ac.uk/
Attachment:
signature.asc
Description: OpenPGP digital signature