#! /usr/bin/perl -w 
use Bio::SeqIO;

#=================================
#Script to rename sequences based on user-defined pattern matching
#briandotoakleyatarsdotusdadotgov
#k.purdy at warwick.ac dot uk
#=================================

#---define input file----
my $usage = "\nUsage:  $0 filename\n\n";
$input_file = $ARGV[0] or die $usage;
chomp $input_file;

#---define output stream----
$out = Bio::SeqIO->new(-fh => \*STDOUT);

my $seq_in  = Bio::SeqIO->new( -format => 'fasta',-file => $input_file);
while ($seq_obj = $seq_in->next_seq()) {
	$id=$seq_obj->id();
	&pattern_match; #Do something with sequence ids;
	$out->write_seq($seq_obj);
}

sub pattern_match {
#EXAMPLES ARE LEFT UNCOMMENTED BELOW.  EDIT TO SUIT.

if ($id =~ m/^HJSJARD01/) {$group=Diseased} #define some group based on regex
if ($id =~ m/^HJJBOJ401/) {$group=Healthy}
$seqid = substr $id, -5; # takes last 5 characters from each id
$new_id = $group."_".$seqid; #makes new id based on groups and seqids
$seq_obj->id($new_id); #resets id for seq object

}