A more "Perl-ish" way of doing the same thing...
#Program to find motif site in a given protein sequence using files
my $motif = "AGGGGG";
open( my $read, "<dna.txt" );
my @e = <$read>;
$_ = join( " ", @e );
push @c, pos( ) - length( $motif ) + 1 while /$motif/g;
s/$motif/color( 'bold green' ) . $motif . color( 'black' )/eg;
print $_, "\n";
print "Number of sites the motif (AGGGGG) is present: ", scalar @c, "\
print "And the positions in the string are: ", join( ',', @c ), "\n\n"
The eliminates counting characters one at a time, as in the $i loop in the original, in favor of using pattern matching on the entire character string. I have found that eliminating loop counters wherever possible greatly reduces the number of bugs in my code.