[Bioperl-l] All-vs-all BLAST

Jason Stajich jason at cgt.duhs.duke.edu
Tue Jul 20 10:52:18 EDT 2004


If you want to run blast I don't think you really need Bioperl.

For a protein seq set called 'sequences'

% formatdb -i sequence -p T -o T
output with -m9 tab output using E < 1e-3
% blastall -p blastp -i sequence -d sequence -o seq-vs-seq.BLASTP -m 9 -e 1e-3

Or in wu-blast
% wu-formatdb -i sequence -p T
(also consider W and T parameters to get some speed depending on
 what you want to find)
% blastp -i sequence -d sequence -o seq-vs-seq.BLASTP -postsw -links E=1e-3 -wordmask seg

Or FASTA
Mask low-complexity first
% pseg sequences -q -z 1 > sequences.pseg
Then run (remove the _t if you don't want threaded) also consider mpi or
pvm version if you have a cluster
% fasta34_t -Q -S -m 9 -d 0 -E 1e-3 sequences.pseg sequences.pseg > seq-vs-seq.FASTA

Bio::SearchIO can parse all these outputs.

-jason
On Tue, 20 Jul 2004, Paulo Almeida wrote:

> Hi,
>
> You want to compare all the possible pairs of sequences in the set? If
> that is it, you can try putting all the sequences in an array, blast the
> first against all the others, shift it out of the array and then repeat
> the process until the array is empty. If it's a multiple alignment you
> want, you could probably use Bio::Tools::Run::Alignment::Clustalw
> <http://doc.bioperl.org/releases/bioperl-1.4/Bio/Tools/Run/Alignment/Clustalw.html>
> (I was going to link to the webpage, but that link is broken in the
> Bioperl webpage).
That would be:
http://doc.bioperl.org/bioperl-run/Bio/Tools/Run/Alignment/Clustalw.html

I didn't bother making a release dir for bioperl-run at this point.
>
> -Paulo
>
> Peter J Stogios wrote:
>
> > Hello,
> >
> > I'm looking for a script to do an all-vs-all BLAST comparison of a set
> > of sequences.
>
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at portal.open-bio.org
> http://portal.open-bio.org/mailman/listinfo/bioperl-l
>

--
Jason Stajich
Duke University
jason at cgt.mc.duke.edu


More information about the Bioperl-l mailing list