# Additional file 2
# to run this program, you need two text files:
# 1. the locus sequence (referred to here as locusfile.txt); this should be a sequence with the repeats in lowercase
# 2. the breakpoint sequences (referred to here as breaks.text); this should be a list of the breakpoints, one on each line;

# to run the program, go to command line, go into the correct directory, and type:
# find_site.pl locusfile.txt breaks.txt


# this part of the program reads locusfile.txt and creates $seq from it
my $locus= $ARGV[0];
open(IN1,"<$locus") or die $!;
my $seq = '';
while (<IN1>){
      chomp;
      $seq = $seq . $_;
};
close (IN1);

# this part reads breaks.txt, matches them to $seq, and returns their distance to a repeat
my $breaks= $ARGV[1];
open(IN2,"<$breaks") or die $!;

open(OUT,">outputfile.txt") or die $!;

$numberofbreaks = 0;

while ($breaksequence=<IN2>){
  $breaksequence =~ s/^\s+//;
  $breaksequence =~ s/\s+$//;
  $rc = scalar reverse $breaksequence;
  $rc =~ tr/ACGTacgt/TGCAtgca/;

  if ($seq =~ m/($breaksequence)/i) {
    my $matchposition = $+[0]-1;
    my $matchletter = substr($seq,$matchposition,1);
    my $distance = getDistance($seq,$matchposition);
    $numberofbreaks = $numberofbreaks +1;

    print OUT "$breaksequence matches\t$matchposition\t$matchletter\t$distance\n";
  }
  elsif ($seq =~ m/($rc)/i) {
    my $rcmatchposition = $-[0];
    my $rcmatchletter = substr($seq,$rcmatchposition,1);
    my $rcdistance = getDistance($seq,$rcmatchposition);
    $numberofbreaks = $numberofbreaks +1;

    print OUT "$breaksequence was reverse complemented to $rc\t$rcmatchposition\t$rcmatchletter\t$rcdistance\n";
  }
  else {print OUT "$breaksequence and its reverse complement have no match";
  };
};

print OUT "\nYou matched $numberofbreaks breaks.\n" ;

# this part does the same thing for 1000 random positions for each breakpoint matched

print OUT "\nrandom site no.\t position in sequence \t distance to a repeat\n" ;
for (my $i = 0; $i <= (1000 * $numberofbreaks); $i++) {
                                my $randdistance;
                                my $seqLen = length($seq);
                                my $randomPos = int(rand($seqLen));
                                my $randdistance = getDistance($seq,$randomPos);
                                print OUT "$i\t$randomPos\t$randdistance\n";
};

close OUT;


# this subroutine measures the distance between a position in $seq and the nearest repeat
sub getDistance {
                # defining variables
                my ($seq,$matchposition) = @_ ;
                my $matchletter = substr($seq,$matchposition,1);
                my $maxD = length($seq);
                my $leftDistance = $matchposition;
                my $rightDistance = length($seq) - $matchposition;

                # defining the distance
                # if the position in question is lowercase, the distance is 0
                if ($matchletter eq lc($matchletter)) {
                                return 0;
                # else search left (1st while) and right (2nd) then return the shortest distance (if, else)
                } else {
                                $i = $matchposition;
                                while (($i >= 0) && ($leftDistance == $matchposition)) {
                                                $i--;
                                                $val = substr($seq,$i,1);
                                                if ($val eq lc($val)) {
                                                                $leftDistance = (abs($matchposition - $i))-1;
                                                }
                                }
                                $i = $matchposition;
                                while ($i < length($seq) && $rightDistance == (length($seq) - $matchposition)) {
                                                $i++;
                                                $val = substr($seq,$i,1);
                                                if ($val eq lc($val)) {
                                                                $rightDistance = abs($matchposition - $i);
                                                }
                                }
                               if ($leftDistance < $rightDistance) {
                                             return $leftDistance;
                                } else {
                                             return $rightDistance;
                                }
               }
}