Forgive the forwarded email. Not used to working from an iPad.  

On 2011-09-29, at 7:58 PM, Peter Cock <p.j.a.cock@googlemail.com> wrote:

On Thu, Sep 29, 2011 at 7:49 PM, Cecilia Tamborindeguy
<ctamborindeguy@ag.tamu.edu> wrote:
Hello,

I would like to know if Galaxy can do de novo assembly without a reference
genome.

Thanks.

Cecilia

Are you trying to use the Public Galaxy or a local install? There
are several assemblers with Galaxy Wrappers on the Galaxy
ToolShed (e.g. Roche "Newbler", and MIRA 3) which you could
add to your own local Galaxy if you have one.

However, do novo genome assembly can be very computationally
demanding, so not many Galaxy Instances will want to offer it.

Peter

I would like to echo Peter's advice. (Again, that's twice in 5 min from 2 different lists. I promise you I'm not stalking you, Peter).

Genome assembly is a bit of a dedicated domain with respect to expertise and time. If possible, if you are assembling a lot of genome data, you really should set yourself up properly with a multiple-CPU unix box with a lot of RAM and dedicate it to assembly. Install MIRA, Newbler, Velvet, AMOS, samtools, bamtools, bedtools, Staden, phred/phrap/consed on it, and you can assemble and interconvert data to your heart's content.

Galaxy is a wonderful and useful service but assembling genomes does require dedicated power and expertise, and preferably in house. I just forked out for a 64 CPU processor with 1 TB RAM bc we assemble lots of genomes. You don't have to go that far but a 3-4 quad processor box with 128 GB RAM and 1 TB disk should be on your mind.

John