[BioC] Computer for the analysis of high-throughput genomic data

[Steve Lianoglou]

Hi,

On Fri, Dec 28, 2012 at 4:36 AM, Capurro, Alberto  (Dr.)
<ac331 at leicester.ac.uk> wrote:
> Thank you very much. I will do microarray analysis at first but in the future we are also interested in sequencing. The computer is for the lab, I will be in charge of the processing, I have experience in computational neuroscience but not in genomics, so I am learning now. I think that the Uni  usually buys windows machines. Regarding the operating system, is there an important reason to use linux instead of windows 7 to run bioconductor and R?. I can use linux if it is better. I can get 10 T and backup in and external disk and in space provided by the Uni network.

Without inciting a flamewar, I don't think it's too controversial to
say that most scientific tools in this space are written for linux
first, then tweaked to run on osx (us osx folks are, by default, stuck
on an older version of gcc, so some tweaks are harder than others),
and likely windows is the after thought.

Look at, for example, some of the aligners out there.

* Bowtie provides compiled binaries for linux and osx, no windows:
  http://sourceforge.net/projects/bowtie-bio/files/bowtie2/2.0.4/

* The STAR aligner runs on linux, and recently was tweaked to run on
osx (not sure if it's entirely working).

* bwa's SF page suggests it only runs on linux and BSD (osx).

* "A unix system" is listed as a prerequisite for installing GSNAP.

For the most part, however, this isn't true for the R/bioconductor
packages you will likely be using. AFAIK, the majority of the bioc
packages work just fine on unix, osx, and windows.

Also, if you're planning on having several people log into the machine
to do work, then I think a *nix is likely going to be your best bet.

So, to be honest, even though I have a slight osx bent, if I were in
your shoes and was put in a position to buy a workhorse machine, I'd
go linux. I assume you, and the other members in the lab, will have
their own desktops/laptops to do downstream analysis -- which can be
the OS of your choosing.

After doing some of the heavy lifting on a compute-server (I'm
thinking of alignment/assembly), you can likely do most all of your
work on a lower powered machine -- especially if we're talking about
more "canned"/routinary analysis. I've done lots of downstream
analysis on my 8gb ram, dual core macbook pro, for instance, although
having access to some big iron to do some heavy computing at times is
totally necessary.

HTH,
-steve

-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
 | Memorial Sloan-Kettering Cancer Center
 | Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact