DNA computing progress

On February 24th, National Geographic reported on progress in DNA computing. This field began in 1994 with Leonard Adelman's work on solving the Hamiltonian path problem, and then I didn't hear much about it after that.

Ehud Shapiro and his colleagues at the Weizmann Institute in Israel have developed a DNA computer that can perform 3.3x10¹⁴ operations per second. A cubic centimeter of "computer soup" contains about 1.5x10¹⁶ individual computers, with a memory capacity of about 6x10²⁰ bytes (ten billion 60-gigabyte hard drives). The two primary advances are that the system is generally programmable, and that the computation is powered by DNA rather than ATP.

"Autonomous bio-molecular computers may be able to work as 'doctors in a cell,' operating inside living cells and sensing anomalies in the host," said Shapiro. "Consulting their programmed medical knowledge, the computers could respond to anomalies by synthesizing and releasing drugs."

This approach appears applicable only to embarrassingly parallel problems involving no inter-processor communication. It should be able to tackle some problems in global optimization, though probably not very complicated ones like de-novo protein design, one of the possible pathways toward advanced nanotechnology. For the time being, it's not yet clear that this is a very useful technique.