[Bioperl-l] motif features?

Shawn shawnh@fugu-sg.org
18 Oct 2002 01:30:09 +0800


Hi all,
	I'm toying around with some motif finders mainly FootPrinter right now.
I was wondering whether there have been any work to represent
motifs in bioperl. Dugged around but I'm not sure that attaching
SeqFeatures to sequence would do the trick.

Mainly what I would like to represent is that given n sequences, we
would have m different motifs (possibly overlapping) found across them.

What I'm thinking of is to have some kinda
Bio::Seq::Motif or Bio::SeqFeature::Motif or Bio::Motif 
that has a bunch of SeqFeatures attached to it representing
where this motif lies on the list of sequences.

A use case:

use Bio::SeqIO;
use Bio::Tools::Run::FootPrinter;

my $sio = Bio::SeqIO->new(-file=>"seq.fa",-format=>"fasta");
my @seq;
while(my $seq = $sio->next_seq){
	push @seq, $seq;
}

my $fp = Bio::Tools::Run::FootPrinter->new();

my @motif = $fp->run(@seq);

foreach my $m (@motif){
 print $m->consensus."\n";
 foreach my $seqfeat($m->seqfeatures){
  print $seqfeat->seq_id."\t".$seqfeat->start."\t".$seqfeat->end."\n";
 }
}

Other methods like

$motif->information_content (bits)
$motif->pssm (weight matrix)
$motif->width
$motif->evalue etc...



Throwing this out for discussion and half hoping someone's already done
something like that but I'm not finding it or that its a worthwhile
expedition.


cheers,

shawn