Could you provide a sample of a subsequence that is supposed to match (there is no TATAAT in this sample)?  Also, may the subsequences in question be split across more than one line? In this case, you would need to chomp the input lines to get rid of the newlines in your $text (as you're not accounting for them in the regex).

Also, if your input is then all on one line, you probably want non-greedy matching ([ACGT]+?) of the parts in between the strings of interest, or else they'll gobble up more than you want...