#!/usr/bin/perl
# repeat_classer2.pl
# usage: rep_classer2.pl rep_units.txt
# run this script before repeatalyzer.pl
# detects what macroclass each repeat belongs to 
# rep_units.txt is a list of all distinct types of repeats in the db 
# Perseus Missirlis - Mar 18, 2004

use strict;

my $rep = "rep_class.txt";

unless ( open(REP_CLASS, ">$rep") ) {
   die "Cannot open file \"$rep\" to write to!\n\n";
      }
	  
my $i = 1;

while (<>) {
  chomp;

if ($_ =~ /(\S+)/) {

    print "line $i matched: $1\n\n";
	
	my @units;
	my $rep_class;
    my @repeat = split('',$1);
    my $lengther = scalar @repeat;
    my $n;

    print "this is the repeat: @repeat\n";
    print "this is the length of the repeat: $lengther\n\n";

	my $x;

    while ($n < $lengther) {
		my $for = shift @repeat;
		$repeat[($lengther - 1)] = $for;
		my $for_run = "@repeat";
		$for_run =~ s/\s//g;
		my $rev_run = reverse $for_run;
		$rev_run =~ tr/ATGCatgc/TACGtacg/;
	
		print "this is the repeat after popping it: @repeat\nthis is the variable for regex: $for_run\nthis is the variable in reverse: $rev_run\n\n";

		$units[$x] = $for_run;
		$x++;
		$units[$x] = $rev_run;
		$x++;
		$n++;
		}

		print "this is the units array: @units\n";

		@units = sort(@units);
		print "sorted units:\n";
		foreach my $unit(@units) {
		print $unit . "\n";
		$rep_class .= $unit . "o";
		}
	$rep_class = "o" . $rep_class;
	print "this is the rep_class: $rep_class\n\n";
	print REP_CLASS $rep_class . "\n";
	$i++;	
	}
}
close(REP_CLASS);
exit;

##############
# End script #
##############