## Script description: This program takes input and match the chromosome location from Bowtie parser file,
#                     then adds the read # to the window (also including neighboring windows)


## Perl interpreter command
# use warnings;
use strict;


## Hashes initialized
my %input;
my %bowtieP;


## Open files


# file1 - input: create window # from position
#my $file1 = shift; # Input (chr # and position - base pair)
open (FILE1, "/data/GAII/prostate_cells/SNP/Prostate_SNP_filtered.txt");


my $head = <FILE1>; # If there is a header in the input #1, include this code
while (<FILE1>) {
    chomp;
    my $input_orig = $_;
    my @line = split /\s+/, $_;
    my $chr = $line[1];
    my $w = (int($line[2]/100))*100;  # Window # created
    my $w1 = $w-100;
    my $w2 = $w+100;
    my $pos = "chr$chr\_$w";       # Get new variable chr_win for input
    $input{$pos}[0] = $input_orig;
    $input{$pos}[1] = $pos;
    $input{$pos}[2] = 0;        # Read value initialized as 0
    $input{$pos}[3] = $chr;
    $input{$pos}[4] = $w;
    $input{$pos}[5] = "chr$chr\_$w1";
    $input{$pos}[6] = "chr$chr\_$w2";
    $input{$pos}[7] = 0;        # Read value for win-100
    $input{$pos}[8] = 0;        # Read value for win+100

}
close FILE1;


# file2 - bowtie parser output
#my $file2 = shift; # Bowtie Parser Input file (chr #, window #, and read value)
open (FILE2, "/data/GAII/prostate_cells/DU145v2_Bow_Per100.txt");

while (<FILE2>) {
    chomp;
    my @line = split /\s+/, $_;
    my $chr = $line[0];
    my $w = $line[1];
    my $read = $line[3];
    my $pos = "$chr\_$w";  # Get new variable chr_win for bowtie parser
    $bowtieP{$pos}[0] = $pos;
    $bowtieP{$pos}[1] = $read;
}
close FILE2;

open (OUT, "> test.txt"); ### Change if file name changes
print OUT "Name\tChr\tPosition\tGenTrain Score\tPrEC\tPrEC alleles\tRWPE\tRWPE alleles\tLNCaP\tLNCaP alleles\tDU145\tDU145 alleles\tchr_win\tread\tchr_win-100\tread-100\tchr_win+100\tread+100\ttotal\n";

foreach my $pos (keys %input) {
	if (exists $bowtieP{$input{$pos}[1]}[0]) {
       	$input{$pos}[2]=$bowtieP{$pos}[1]; # Change read value from input (initialized as 0) to the read value from bowtie parser data
           }
	if (exists $bowtieP{$input{$pos}[5]}[1]) {
            $input{$pos}[7] = $bowtieP{$input{$pos}[5]}[1]; # Change read value from input (initialized as 0) to the read value from bowtie parser data
        }
	if (exists $bowtieP{$input{$pos}[6]}[1]) {
            $input{$pos}[8] = $bowtieP{$input{$pos}[6]}[1]; # Change read value from input (initialized as 0) to the read value from bowtie parser data
        }
	 my $total = $input{$pos}[2] + $input{$pos}[7] + $input{$pos}[8];
   	print OUT "$input{$pos}[0]\t$input{$pos}[1]\t$input{$pos}[2]\t$input{$pos}[5]\t$input{$pos}[7]\t$input{$pos}[6]\t$input{$pos}[8]\t$total\n";
}

close OUT;
exit;