There's a literature on encoding data into DNA sequences, which doesn't seem to be cited much. That worries me.

Second, they don't exploit any of the real vulnerabilities they found. I'd like to know why. I think the reasons would be interesting.