#file: PWMScore.pl
#input: text file of 19 nucleotide sequence strings
#output: calculates matrix similarity score for input sequence strings compared to position weight matrix of bona fide dioxin response elements

$file1 = shift;
open FILE1, "<$file1" or die "cannot open file1:$!";

@ci = qw(34 33 15 15 24 33 55 100 100 100 100 100 57 38 37 25 37 23 21); #defines the array of Ci vectors

$pwm{1} = [7, 53, 33, 7];	#defines hash of arrays containing nucleotide distribution frequencies [A, C, G, T] for each position {1-19}
$pwm{2} = [33, 40, 27, 0];
$pwm{3} = [27, 20, 33, 20];
$pwm{4} = [20, 27, 33, 20];
$pwm{5} = [7, 40, 20, 27];
$pwm{6} = [7, 13, 20, 60];
$pwm{7} = [2, 7, 0, 73];
$pwm{8} = [0, 0, 100, 0];
$pwm{9} = [0, 100, 0, 0];
$pwm{10} = [0, 0, 100, 0];
$pwm{11} = [0, 0, 0, 100];
$pwm{12} = [0, 0, 100, 0];
$pwm{13} = [47, 53, 0, 0];
$pwm{14} = [13, 40, 47, 0];
$pwm{15} = [60, 27, 7, 7];
$pwm{16} = [47, 20, 27, 7];
$pwm{17} = [20, 27, 53, 0];
$pwm{18} = [33, 40, 20, 7];
$pwm{19} = [7, 27, 33, 33];

open OUT, ">>matrixscore.txt";	#opens file for output
while (<FILE1>) {			#the meat and potato(e)s
	chomp;
	@seq = $_ =~ /\w{1}/g;	
	$i=1;
	foreach (@seq) {
		if ($_ eq a) 	{$score = $score+(($pwm{$i}[0]) * ($ci[($i-1)])) }
		elsif ($_ eq c) 	{$score = $score+(($pwm{$i}[1]) * ($ci[($i-1)])) }
		elsif ($_ eq g) 	{$score = $score+(($pwm{$i}[2]) * ($ci[($i-1)])) }
		elsif ($_ eq t) 	{$score = $score+(($pwm{$i}[3]) * ($ci[($i-1)])) }
		$i++;
		}	
	$matrix_similarity_score = $score / 72858;
	print OUT "$matrix_similarity_score\t$_\n";
	$score = 0;
		}