What I really need is a mathematician help me out a bit in advance, before I get any results. My computer is now crunching at 500 million base pairs. If generated at random, there is a chance of one in 26^100 of getting a certain English sentence 100 characters in length. That number is astronomically small, there are only 10^80 or so atoms in the entire universe. But with the way I'm searching the DNA, what is the probability of finding a sequence 100 characters in length? The math is beyond my abilities. So far I've got 42 characters for Lorem Ipsum and 45 for the hebrew Bible. So the results so far do not in any way suggest a divine author for DNA. But what result would? What would be needed to convince people? 100 characters, 5000 characters or more?

To everyone saying that with this method you can find the entire works of Shakespeare or the communist manifesto embedded in DNA I say - give it a try. Use my script or better yet, write your own. It's not possible, not with the encoding I'm using.

Nobody has spotted the obvious, that the code for reading the file is painfully slow.

A significant improvement would be:

while (read($fh, my $char, 1) && $eof) {
    
    if ($char =~ m/[ACGT]/ ) {
        $i++;

        if ($i >= $start_genome && $i < $end_genome) {
            push (@genome, $char);        
        } elsif ($i > $end_genome) {
            $eof = 0;
                # do the searching here, instead of outside the loop!
        }
        
   }
}
[download]

That's now fixed and the algorithm doesn't go back to the beginning of the file every time a chunk is read. Silly me.

So far I've got 42 characters for Lorem Ipsum and 45 for the hebrew Bible.

Sorry, can't help you more right now. See Multiple comparisons problem. Any good statistics handbook should explain it in depth.