instead of calling index repeatedly.

The tradeoff is:

• scanning a 4 character string for a single character.
• hashing a single character to a 32-bit hash and then performing a modulo 4 operation upon it.

Which actually favours the former:

#! perl -slw
use strict;
use Benchmark qw[ cmpthese ];

our %lookup = ( A=>0, B=>1, C=>2, D=>3 );
our $input = 'ACGT' x 1000;

cmpthese -1, {
    index => q[
        our( %lookup, $input );
        my $n;
        $n = index "ACGT", $_ for split '', $input;
    ],
    hash  => q[
        our( %lookup, $input );
        my $n;
        $n = $lookup{ $_ } for split '', $input;
    ],
};

__END__
c:\test>junk5
       Rate  hash index
hash  107/s    --  -21%
index 135/s   27%    --
[download]