#!/usr/bin/perl
use warnings;
use strict;
use v5.10;

#  Create program to read in a series of VCF files, outputting a two-column
#  file consisting of chromosome and coordinate for each site seen in one or
#  more files ('master-site-list').

die "Give the names of each VCF" if @ARGV < 1;

my $number_of_files = @ARGV;
my $infile;
my $current_file_number = 0;

my %hash_of_chroms;

open my $outfile, '>', "master-site-list" or die "Cannot open output file: $!";

until ( $current_file_number >= $number_of_files ) {

    open $infile, '<', $ARGV[$current_file_number]
      or die "Cannot open VCF file: $!";

    while (<$infile>) {
        chomp;
        next if /^#/;

        my ( $chr, $coord ) = split(/\s+/);

        push @{ $hash_of_chroms{$chr} }, $coord
          unless defined $hash_of_chroms{$chr}[$coord];   # don't add duplicates
    }

    $current_file_number++;

    close $infile;

}

my %sorted_hash_of_chroms;
foreach my $chrom ( keys %hash_of_chroms ) {
    push @{ $sorted_hash_of_chroms{$chrom} },
      sort { $hash_of_chroms{$a}->[0] <=> $hash_of_chroms{$b}->[0] }
      keys %hash_of_chroms;
}

foreach my $chrom ( keys %hash_of_chroms ) {
    say $outfile "$chrom\t@{ $hash_of_chroms{$chrom} }";
}

close $outfile;