Kuo Lab Wiki

# count_EcoRI_site.pl
# Read a fasta file and count the number of 
# EcoRI restriction sites in each sequence. 
my $in_file = shift;
 
my $seq_hash; # key = seq_name, value = seq;
{
	# redefine the record separator
	local $/ = ">";
	open IN, "<$in_file";
	my $in_line = <IN>; # toss the first record
	while ( $in_line = <IN> ) {
		chomp $in_line; # remove the ">" character in the end 
		my ( $seq_name, $seq ) = split( /\n/, $in_line, 2 );
		$seq =~ tr/ \t\n\r//d;    # Remove whitespace
		$seq_hash{$seq_name} = uc $seq;
	}
	close IN;
}
 
foreach my $seq_name ( sort keys %seq_hash ) {
	my @sites = $seq_hash{$seq_name} =~ /(GAATTC)/g;	
	print "$seq_name has ",
	  scalar @sites, 
	  " EcoRI sites\n";
}