#!/usr/local/perl
#we modified the main program by removing processing the likelihoods so we get faster program to count dif in two individuals from same population
$chunk =$ARGV[1]*1000000 +1;

$dir = 'PUR_Rare_April28';
mkdir "./$dir", 0750 unless -d "./$dir";

open(OUT1, ">./$dir/$ARGV[0]_$ARGV[1]_Aa")|| die "Can't open 1_1 output : $!\n";
#	open(OUT2, ">./$dir/$ARGV[0]_$ARGV[2]_$ARGV[3]_B")|| die "Can't open 2_2 output : $!\n";

$str = localtime();
#print OUT1 "$str\n";
#@pair_id = ();
open (FILE1, "zcat /home/ahmed/PUR_RareSNPs_Table.txt.gz |") || die "Can't open RareSNPs table file  : $!\n";
$line_N = 0;
while (<FILE1>){
	$line_N++;
	if ($line_N == 1){
		chop $_;	
		@ID = split(/\t/, $_);
		$n = 1;
# to test that we are getting the IDs line 
		foreach (@ID) {
 			print "$n\t$_\n";$n++;
 		} 
		last;
	}
}

#new edit to test for two individuals.
#@pair_id = ('NA19328', 'NA19312');#TWo GBR individual to test

for $x (0..$#ID) {
        for $y (0..$#ID) {
                        if ($ID[$x] eq $ID[$y]) {    push(@columns, $x);
#                        #end of the edit.
                        }
         }
}
for $n(0..$#columns){
print $columns[$n]."\n";}
#

#
open (FILE, "zcat /home/ahmed/PUR_RareSNPs_Table.txt.gz |") || die "Can't open zipped file chr21 : $!\n";
$line_N = 0;
#$w, $z two individuals from the analyzed population(e.g. GBR)
@diff = ();
while (<FILE>){
	$line_N++;
	if ($line_N >= $chunk + 1000000) {last;}
	if ($line_N > 1 && $line_N >= $chunk) {

	if ($line_N%1000 ==0   ) {print "$line_N PUR\n";}
	if ($line_N > 1){
#		print "my god\n";
		@e = split(/\t/, $_);
#		print $e[1]."\n";
		for $w (0..$#columns){
			for $z (0..$#columns){
				if ($w < $z){
				#	@e = split(/\t/, $_);
				#	1.To calculate Differneces between each two individuals.
					if ($e[$columns[$w]] =~/^(\d)\|(\d)\:(.+)\:(.+)\,(.+)\,(.+)/){
					$p1 = $1 + $2;
		#			print $p1."\n";	
					}
					if ($e[$columns[$z]] =~/^(\d)\|(\d)\:(.+)\:(.+)\,(.+)\,(.+)/){
                                        $p2 = $1 + $2;
	#				print $p2."\n";
                                        }
					#to calcualte total differences 
				#	$diff[$w][$z] += abs ($p1 - $p2);
				#	$my = abs ($p1 - $p2);
#					print $my."\n";
#					End of 1.
#					to calcualte shared rare SNPS.
					if ($p1 == 1 && $p2 == 1){$one_one_common[$w][$z]++; 
					#print "found common \n";
					}

					if ($p1 == 2 && $p2 == 2){$two_two_common[$w][$z]++;}
					if (($p1 == 2 && $p2 == 1) || ($p1 == 1 && $p2 == 2)){$two_one_common[$w][$z]++;}
					
				}
			}
		}
	
	}
	}
}
for $w (0..$#columns) {
         for $z (0..$#columns) {
                if ($w < $z) {
 		#to print total differences : 
 		#	print  OUT1 "$columns[$w]\t$columns[$z]\t$diff[$w][$z]\n";
 		# to Print Shared rare SNPs.:
 		$Total = $one_one_common[$w][$z] + ($two_two_common[$w][$z]* 2) + ($two_one_common[$w][$z]);
 		print OUT1 "$columns[$w]\t$ID[$columns[$w]]\t$columns[$z]\t$ID[$columns[$z]]\tOne_One:\t$one_one_common[$w][$z]\tTwo_Two:\t$two_two_common[$w][$z]\tTwo_One:\t$two_one_common[$w][$z]\t $Total\n";

         	}
         } 
}
$str2 = localtime();
print OUT1 "$str2\n";