Aw: Re: Bioinformatics course in debian-med - feasibility?
Hi David,
> I very much like the idea of helping debian-med on ARM in some way. Some
> of our 4273pi Bioinformatics for Biologists exercises could be scripted,
> so presumably would work as regression tests. We often don't know the
> 'real' result but can get the result on AMD64, and check for deviation
> from that.
>
> Entirely agreed: the Raspberry Pi is not powerful. Somewhere between a
> Cray 1 and a Cray 2, and with 512 GB RAM. Tony's Parallella will be
M
> enormously more powerful and will be great to see.
Indeed. I _almost_ supported the kickstarted campaign at the time and
now wait for the successor myself.
> Our teaching on Pi uses full 'industrial-strength' analyses, but on small
> input. We have not yet used the Pi for genome assembly. Others have:
>
> https://twitter.com/richardmleggett/status/211944292128796672
> http://hal.inria.fr/hal-00842027
Following :)
> The 4273pi Bioinformatics for Biologists course does not currently use
> strap. (It uses Clustal-X, not for aligning but for previewing a multiple
> alignment generated by MAFFT.) We have someone lined up to improve
> coverage of multiple alignment in the future - I'm not sure what he'll use.
I am not sure that it matters what exactly you are covering in the course.
It should just be somewhat complete from a wet-lab's raw data to a
publication. I can be of some help here if required.
> Agreed: At this stage we do not have to be extremely detailed about what
> distribution to contribute to. I am happy to leave this somewhat open for
> now.
"distribution" is Linux distribution? When you mention "contribution", I
would like to see distinguished
a) a contribution to the Debian Med SVN/GIT source code repositories
b) a contribution of a package to some distro
To me, a) is the important part. Some good soul needs to upload to
Debian to see packages migrating from there to Ubuntu from which again
BioLinux takes them or performs its own adoption independently. I do
not have an exact idea about where the Raspberry fits in in above chain,
admittedly. I am just fairly confident that we should not introduce
to much of a deviation/shortcut from regular workflow to get packages in
there - whatever this is. We just all get too confused, otherwise.
> Regarding BioLinux possibilities: Ubuntu does not run on the Pi (ARM11 /
> ARM v6). It does run on the somewhat faster and more expensive Cotton
> Candy (Cortex-A9, ARM v7,
> http://www.fxitech.com/cotton-candy/what-is-it/). I will soon own one of
> these. But irrespective of all this, we absolutely want our future
> teaching material to be able to 'slot into' BioLinux and will make sure
> that it fits.
I have an Android version also on my TV - from what I understood this
has no IO, unlike the Pi which was designed for kids solderin their bits
to it. Fine for most use cases. I suggest to instead give the Pi-people
a kick to give some extra beef to their beast.
Sorry for all this UK networking, but please contact
http://www.linkedin.com/in/garrettmorris who I know to have used the
Pi before. Any such link to drug discovery should help your application
considerably.
Finally, the BOINC people have support for Android phone and there will
soon be support for the Raspberry. Please consider also contacting
http://boinc.berkeley.edu/anderson/ and for respective advice about how
your teaching and grant application fits in.
Best,
Steffen
Reply to: