This is pretty nifty from SciAm.
AND Gates on DNA. The process looks very involved but the new key I see here is that they can feed output directly into another calculation.
I imagine this could lend itself well to massively parallel computing.
Squeeze a few drops into a bunch of test tubes apply the filter gates for the thresholds. Wait for a few hours then sequence the output to find the next Mersenne prime. Diffie-Hellman look out.
I wonder what the key limiting points are for this.
The other interesting item was the manipulators.
The Singularity approaches.