was using the name $role but in the code below I was using $key instead of $role

I changed this code

for my $href ( @AoH ) {
  for my $role ( sort keys %$href ) {
                     my $key;
    if ( ! exists $sequences{$key} ) {
        warn "KEY $key in File1 not found in File2\n";
        next;
    }
    if ( length( $sequences{$key} ) < $href->{$role} ) {
        warn "KEY $key: File2 string length too short for File1 positi
+on value\n";
        next;
    }
    my $index = rindex( $sequences{$key}, "TTT", $href->{$role} );
    if ( $index < 0 ) {
        warn sprintf( "KEY %s: No TTT in File2 string prior to positio
+n %d\n",
                      $key, $href->{$role} );
        next;
    }
    $index += 3 while ( ($index + 3) < $href->{$role} );
    print "$key $href->{$role} " . substr($sequences{$key}, $index, 3)
+ . "\n";
}
[download]

with this code

for my $href ( @AoH ) {
  for my $key ( sort keys %$href ){
    if ( ! exists $sequences{$key} ) {
        warn "KEY $key in File1 not found in File2\n";
        next;
    }
    if ( length( $sequences{$key} ) < $href->{$key} ) {
        warn "KEY $key: File2 string length too short for File1 positi
+on value\n";
        next;
    }
    my $index = rindex( $sequences{$key}, "TTT", $href->{$key} );
    if ( $index < 0 ) {
        warn sprintf( "KEY %s: No TTT in File2 string prior to positio
+n %d\n",
                      $key, $href->{$key} );
        next;
    }
    $index += 3 while ( ($index + 3) < $href->{$key} );
    print "$key $href->{$key} " . substr($sequences{$key}, $index, 3) 
+. "\n";
}
[download]

I am getting exact outputs now ;)

#!/usr/bin/perl

use warnings;
use strict;
use Data::Dumper;

my $fh1 = "File1.txt";
my $fh2 = "File2.txt";

#> so far, so good. though in a flexible program at least better writt
+en as
#> that way you can call it: perl scriptname file1 file2
# my ($position_file, $sequence_file) = @ARGV;

open(FH, "$fh1") || die ("Can't open: $!");
my @AoH;

#> that's "ok", too. three quirks: - unnecessary quoting. if you mean 
+the plain value of a variable
#> there's no need to quote it ("") - The use of a Filehandle. Best pr
+actice are lexical variables for filehandles.
#> - variable names: for sake of you own sanity: use DESCRIPTIVE varia
+ble names. ALWAYS.
#> so:
# open(my $file, '<', $position_file) || die ("Can't open $position_fi
+le: $!");
# my %positions;

while ( <FH> ) {
    push @AoH, { split /[\s]+/ };
}
    # this will not do what you think. it creates a hashref from the a
+rray split of your positions file
    # this will trigger the "Odd number of elements in anonymous hash 
+at ashnator.pl line 12, <FH> line 1." warnings
    # you want:
    
# Create a hash of arrays of positions
# while ( <$file> ) {
#     my ($key, $pos) = split /\s+/;
#     if (!$positions{$key}) {
#         #create array at $key for the first position per key
#         $positions{$key} = [$pos];
#     } else {
#         # add position for all subsequent lines
#         my $ref = $positions{$key};
#         push @$ref,$pos;
#     }
# }

my %sequences;
{ #> uneccessary block
   open(FD, "$fh2") || die("Cannot open: $!");

   my $key;
   while (<FD>) {
       if ( s/^>// ) {
           $key = ( split /\|/ )[1];
       }
       else {
           chomp;
           $sequences{$key} .= $_;
       }
   }
}

#> print Dumper(\%sequences);

#> same critics than above, but this time the loop works and i assume 
+you understand how.

for my $href ( @AoH ) {
    for my $role ( sort keys %$href ) {
         my $key; # why $key when $role is the loop variable? 
         
#> RULE OF THUMB: Think before you copy and paste!
#> because the data structure is now a bit different we make this:
#> while ( my ($key, $position_list) = each %positions )
#> {
#>     for my $position ( @$position_list )
#>     {
#>        print "key: $key, position: $position\n";
#>        # ... your inner loop code
    }
}
[download]

for my $role (sort keys %$href ){
  my $key;
  if (!exists $sequences{$key}){ # line 33
[download]

12345 21 GCT
12345 42 CAG
67891 32 GCA
[download]