Computers that run on the blueprint of life, DNA, rather than conventional computer chips, could be faster than ordinary computers within a decade, scientists told the conference. They have been exploring the potential of DNA for less than a year, after South California scientist Len Adleman suggested that it could be used for this purpose.
DNA is the storage medium holding all the information required to construct a living thing. Scientists have calculated that a kilogram of DNA in a computer could work at a speed of 100 tera-operations per second. Electronic computers are also approaching tera-op speed, said Richard Lipton, professor of computer science at Princeton University. But they require a space the size of a large room to do so.
DNA could store 1017 bits per litre, compared with 1012 bits per litre for magnetic data. DNA contains strands made up of four bases. As the sequence of these bases changes, so does the information it holds. Each DNA strand could be viewed as a little processor. The scientists think they would process the information using biologists' technology: for example, the polymerase chain reaction, in which a scientist can replicate all the strands of DNA in a test tube; or gel electrophoresis, which enables scientists to separate DNA strands according to their lengths.
Professor Lipton's vision of a DNA computer is "a lot of pipes and vats which are bubbling away". But the scientists were a bit subdued because of the major obstacles they face. Dan Boneh, also a computer scientist at Princeton University, said: "Your wrist-watches won't have DNA in them." The main problem is that with DNA, each operation (the equivalent on a chip of a binary or gate operation) would take an hour, because it involves processes such as extraction or mixing. To make up for this the computer would hold many DNA strands operating in parallel.
"The two types of machine are exact opposites of the spectrum," said Dr Boneh. Whereas a conventional computer might contain 100 processors (chips) each of which can do 1011 operations an hour, a DNA computer would contain 1017 processors (strands of DNA) but each one could only do one operation an hour.
The real strength of DNA computing will probably lie in operations that require large storage capacity, they said. An example would be large searches, such as cryptography.