One quick note: since you only have one header line, and you know it's the first line, there's no need to check for it every time through your loop. Read the header line, process it, and then start your loop on the second line.

I'll assume that you really do want a HoAoA, despite the large empty areas it may have within it. In that case, I'd start by reading that header line and making an array of two-element arrays, the first element containing the first part (here 162), and the second element containing the remainder (87215 for the first one). Then as you go through the rest of the lines, you can easily step through this array to get the indexes you'll need for your HoAoA. When you start having indexes within indexes, that's when it gets mind-stretching, but fun. Something like this:

#!/usr/bin/env perl
use Modern::Perl;

my %data;                 # HoAoA holding final data structure
my @indexes;              # array of indexes into %data based on coord
+inates

my $header_line = <DATA>; # some of these steps could be combined;
chomp $header_line;       # including them for clarity
my @headers = split ',', $header_line;
shift @headers; # throw away SAMPLE
for my $n (@headers){
    my $x = $n/100_000;
    my $y = $n % 100_000;
    push @indexes, [ $x, $y ];
}

while(<DATA>){
    chomp;
    my( $id, @bits ) = split ',';
    for my $i (0..@bits-1){
        $data{$id}[$indexes[$i][0]][$indexes[$i][1]] = $bits[$i];
    }
}
say "Expecting 0 0: ", $data{HG00553}[162][87365];
say "Expecting 1 1: ", $data{HG00638}[162][87851];

__DATA__
SAMPLE,16287215,16287226,16287365,16287649,16287784,16287851,16287912
HG00553,0 0,0 0,0 0,0 0,0 0,0 0,0 0
HG00554,0 0,0 0,0 0,0 0,0 0,0 0,0 0
HG00637,0 0,0 0,0 0,0 0,0 0,0 0,0 0
HG00638,0 0,0 0,0 0,0 0,0 0,1 1,0 0
HG00640,0 0,0 0,0 0,0 0,0 0,1 1,0 0
[download]