#Fotiou_Process.pl

use strict;
use warnings;
use Cwd qw(getcwd);

my $start_run = time();
my $row;
my %data;
chdir "/PATH-TO-DIRECTORY";
my $file =  "/Fotiou_complete/Fotiou_Raw_Data/Fotiou_Data_1.csv";
 

#HGNC Gene names 
open my $in, "<", $file or die;
while ($row = <$in>) { 
    foreach ($row) {
        chomp($row);
       	if ($row =~ /^([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^.]*)$/){ 
        	$data{$2}{"$2,$3,$4,$5,$6,$7,$8,$9"} = "";
        	} 
		}
	}
close $in or die;


my @files = glob("/Fotiou_complete/Fotiou_Raw_Data/*.txt");
foreach my $file (@files) {
	my $name;
	my %eliza;
	my $title;
	my $count =0;
	if ($file =~ /\/Fotiou_complete\/Fotiou_Raw_Data\/([^.]*)/i){
		$name = $1; print "$name\n";
		}
	open my $fh, "<", $file or die;
	while ($row = <$fh>) { 
    	foreach ($row) {
        	chomp($row);
        	if ($row !~ /^ENSG/){
        		$title = $row;
        		if ($title !~ /\t$/){
        			$title = $title."\t";
        			}
        		}
        	if ($row =~ /^([^\t]*)([\w\W\d\D\s\S]*)$/g){
        		my $gene = $1;
        		my $info = $2;
        		if ($info !~ /\t$/){
        			$info = $info."\t";
        			}
        		$info =~ s/,/|/g;
        		$info =~ s/\t/,/g;
        		$title =~ s/\t/,/g;
        		$eliza{$gene}{$info} = "";
        		}
        	else{
        		print "X $row\n";
        		}
        	}
        }
    open (my $out, ">", "/Fotiou_complete/Fotiou_Output/$name\.csv")or die "Couldn't open: $!";
    print $out "$title";
    print $out "Haplosufficiency.rank,FFS,Branch.length,Paralog.status,SSD.Partners,Strict.Partners,Intermediate.Partners,Relaxed.Partners\n";
    open (my $out2, ">", "/Fotiou_complete/Fotiou_Output/MISSING_$name\.csv")or die "Couldn't open: $!";
    open (my $out3, ">", "/Fotiou_complete/Fotiou_Output/Hap-Par_$name\.csv")or die "Couldn't open: $!";
    foreach my $gene (sort keys %eliza){ 
		if (exists($data{$gene})) {
			print $out "$gene";
			if (exists($eliza{$gene})) {
                my $count = 0;
				foreach my $k (sort keys %{$eliza{$gene}}){
                    $count++;
                    if ($count == 1) {
                        print $out "$k";
                        if ($k !~ /([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),/){
                            print $out ",";
                            }
                        }
                    else {
                        if ($k !~ /^$/) {
                            print $out3 "$gene";
                            print $out3 "$k";
                            if ($k !~ /([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),([^,]*),/){
                                print $out3 ",";
                                }
                            print $out3 "\n";
                            }
						}
					}
				}
			else{
				print $out ",";
				}
			foreach my $l (sort keys %{$data{$gene}}){
				print $out "$l\n";
				}
			}
		else {
			if ($gene =~ /^ENSG/){
				print $out2 "$gene\n"; #No associated gene found
				}
			}
    	}
    }

exit;