[Bioclusters] Re: wired bioclusters article is now online

[Eric Engelhard]

xpscreens wrote:
> 
> Hi,
>     I was just wondering.. after you've got clusters set up, what software
> do you use, for in this case, to engineer a honeybee.
> 

In short, pair-wise sequence homology searching (BLAST, FASTA) and
protein domain comparisons (InterPro), but it really depends on the
project goal. I am interested in workbench/annotation tools, but only
the homology tools would actually be clustered for performance. The
single most valuable benefit of a cluster (to me, anyway) is the speed
of large batch runs. I write Perl wrappers around everything.

The "venomless" project mentioned in the article involves a technique
called genetic ablation, which is the expression of a cell killing
product under the control of a tightly regulated, tissue specific
promoter. For the honeybee venom gland, three protein venom components
(hyaluronidase, mellitin, phospholipase A) are well understood, so there
was little computational biology involved. The broader goal, however, is
to leverage public domain sequence information from other arthropods
(Drosophila is by far the best model: see http://www.fruitfly.org/) to
guide future efforts. Again, we get back to batched pair-wise homology
and protein domain searches.

Again, my recommendation is to not let the tail wag the dog. Build a
cluster if you really need it, but be aware that you may not need high
performance. Of course, you CAN use a small cluster to rip and encode
music or run PVM/MPI povray. :-)

--
Eric Engelhard - www.cvbig.org - www.sagresdiscovery.com