If I understand your description of the task, it looks like your two snippets of sample data are different in ways that they shouldn't be -- that's a bit confusing.

But based on your description (and assuming that the input data is correctly sorted according to the values of the "reference_position_begin" and "reference_position_end" columns), I think the following would do what you want:

#!/usr/local/bin/perl

use strict;
use warnings;

my $prev_key = '';
my $match_count;

while (<>) {
    my @flds = split;
    my $key = join ' ', @flds[6,7];
    $match_count = 0 if ( $key ne $prev_key );
    $prev_key = $key;

    splice( @flds, 6, 0, ++$match_count );
    print join( "\t", @flds ), "\n";
}
[download]

If the input data isn't sorted properly yet, just look up the unix "sort" command to take care of that.