#!/usr/bin/perl
use warnings;
use strict;
use v5.14;

use Getopt::Long;
use Bio::PopGen::IO;
use Bio::PopGen::Statistics;

die "need two arguments (i.e. chr cont) at invocation" unless @ARGV == 2;

chomp( my $chr_num = shift );
chomp( my $cont    = shift );

open my $out_file, ">", "chr${chr_num}_exome_snps_processed_${cont}_STATS"
  or die "Can't open output file: $!\n";

open my $in_file, "<", "chr${chr_num}_exome_snps_processed_$cont"
  or die "Can't open input file: $!\n";

my %data;
my @snp_bins;
my @individuals;
my @all_snps;

while (<$in_file>) {
    chomp;
    if (/^SAMPLE/) {
        my ( $placeholder, @coords ) = split /,/;
        foreach my $coord (@coords) {
            push @snp_bins, int( $coord / 100_000 );
        }
    }
    else {
        my ( $id, @snps ) = split /,/;
        push @individuals, $id;
		push @all_snps[$. - 2], join(',', @snps);
    }
}

foreach my $individual (@individuals) {
    foreach my $index ( 0 .. $#snp_bins ) {
        push( @{ $data{$individual}[ $snp_bins[$index] ] }, $all_snps[$index] );
    }
}

close $in_file;