"Computing with DNA," by Leonard M. Adleman. Scientific American, August 1998 .
Some of the most complex and intricate computations do not occur in agovernment supercomputer--they are happening inside the nuclei ofcells. Part of the Watson-Crick model of DNA is an amazing littlenanomachine called DNA polymerase, a single molecule that reads aDNA strand and transfers the information to a new, growing strand.This is how genetic information is transferred when a cell reproduces.
Computer scientist Leonard Adleman noticed that this process is similar to the Turing machine, a theoretical device that predates the computer. A Turing machine consists of a pair of tapes and a mechanism which reads data from one tape while simultaneously moving along the other, reading and writing other data. Its similarities with DNA is clear,but what makes it striking is that Turing's machine can compute anything that is computable at all. Adleman theorized that DNAcould likewise be programmed to solve problems.
An experiment was necessary. Adleman chose a version of the famous Hamiltonian path problem, hypothesizing that it could be solved using DNA.He first translated the assumptions of the problem into DNA sequences,then generated substances with this genetic code, and let the substancessynthesize. In about a second, the solution to the problem was created,in terms of newly formed molecules with the appropriate genetic code. From here, Adleman had to separate this code from the hundreds of millions of other codes produced, a process which took several days.
Despite the lengthy time scale, the idea has many attractive properties. DNA provides dense information storage, and because millions of molecules synthesize at once in a very small space, it is capable of enormous parallelism. But beyond this practical application, Adleman's DNA computing experiment represents an emerging field of science made possible by our developing ability to control the molecular world.