Computing with DNA

Fragments of DNA can be used to represent computer data.

Within the area of organic molecular electronics, DNA computing is an emerging interdisciplinary field that unites computer science and molecular biology. Scientists create fragments of DNA, whose letters represent computer data and instructions, and mix them together in test tubes to solve problems, such as the shortest path through a number of cities. At the APS Centennial Meeting in Atlanta, several researchers reported on recent advances that could one day enable the production of a DNA-based electronic device.

Allen Mills of Bell Labs/Lucent Technologies believes it is possible to use DNA to construct a massive neural network — computers modeled after the human nervous system — with a connectivity of 1 trillion synapses, still a mere 1% of the capacity of the human brain. According to Mills, hybridization of pairs of complementary DNA strands makes possible a representation of highly parallel selective operations that he believes could be one key to achieving molecular computation. However, significant undesired pairings of strands frequently result from small departures from the ideal selectivity of DNA hybridization, making it difficult to implement the large-scale Boolean operations necessary for DNA-based computing. Neural networks, which do not require the same high precision as digital computing, offer a promising solution, especially when combined with new rapid techniques for the interconversion of digital and DNA data. Mills' design could also be configured in more general architectures for solving problems of prediction and classification.

Adopting a loftier, theoretical approach to the issue, Simon Berkovich of George Washington University speculates that DNA in a biological organism serves a role comparable to a barcode: it is a pseudo-random number (PRN) that provides classification, so that small differences are enough to distinguish between species. It also provides a unique ID number that is responsible for the biological individuality of an organism. "With this hypothesis, the essence of the phenomenon of Life unfolds as extracorporeal information processing in the infrastructure underlying the physical world," says Berkovich. The corresponding PRN enables biological cells to interact through this infrastructure by sharing its storage and bandwidth resources in a Code Division Multiple Access (CDMA) mode. According to Berkovich, the differences in the behavior of dead and living matter results from differences in the sizes of molecules involved, which affect PRN lengths. A short PRN can pick up only noisy background, while a lengthy PRN can sustain a robust information exchange. Thus, "The macromolecules of the DNA can serve as CDMA transreceivers for interaction with the infrastructure of the physics world, and as microtransducers materializing the control signals into purposeful biological events," he says.