Scientists at Pacific Northwest National Laboratory (PNNL) reported in the journal Communications of the ACM that for very long-term storage of data DNA is the medium of choice. While electronic memory degrades, paper disintegrates, and even stone tablets erode, DNA can store data effectively for millions of years.
DNA has been used as a storage mechanism before, when the Mount Sinai School of Medicine embedded information in a strand of DNA and hid it behind a period in a document, which they then mailed to themselves. They successfully retrieved the information, but naked DNA is easily destroyed - hence the PNNL deciding to try encoding information stored in a living organism.
The organism chosen to store the DNA was Deinococcus radiodurans, a bacterium whose entire genome has been sequenced. Deinococcus is an extremophile (an organism that can withstand extreme conditions), and can survive ultraviolet radiation, desiccation, high temperatures, partial vacuum and ionising radiation at levels 1000 times more than that required to kill a human.
The other benefit of using Deinococcus is the small size required to store the information - one millilitre of water can hold up to one billion bacteria. The scientists point out that different sequences can be inserted into different bacteria, increasing the storage capacity enormously.
DNA consists of four base pairs - Adenine (A), Cytosine (C), Guanine (G) and Thymine (T) - which are arranged in triplets of any combination, called codons. The scientists used this system to assign a different letter, number or signal to each codon, meaning three base pairs can contain one of 64 different bits of information.
Using the binary system three digits can indicate up to seven different bits of information - to get the equivalent information capability as a DNA codon seven digits are required. The scientists managed to store and retrieve DNA 57-99 codons long, equivalent to around 18 Mb.
The information-bearing DNA was stored between two 20-base pair long sequences that are used as markers, and the whole strand inserted into a cloning vector - a circular DNA molecule that self-replicates within a bacterium. This was then inserted into bacteria for long- term storage.
To extract the information the DNA is run through a laboratory procedure called a polymerase chain reaction, which can recognise the marker sequences and increase the number of DNA strands. Analysis can then determine the sequence of the DNA, allowing the information to be read. The whole process of decoding takes about two hours.
And here lies the reason the procedure will never replace electronic storage - encoding and decoding the information takes too long for most processes today. The scientists at PNNL envisage the system being used for the long-term storage of critical data, especially when a mode of storage that can survive a nuclear catastrophe is required.
The other problem with storing information as DNA is the possibility of mutation. Although bacteria can be selected which have a very low mutation rate, random changes still occur.
In similar news, a team from the Weizmann Institute of Science in Israel have published a paper in the Proceedings of the National Academy of Sciences claiming they have created a computer that can perform 66 billion transitions per microlitre per second. What's really special about this computer is that the DNA used to input the information also inputs the energy required to perform the calculation.
In this case, the computer uses DNA as software and enzymes as hardware. While the computer is very small and very fast, it can only perform a few types of computation at the moment, such as determining whether a string of information containing "a" and "b" has an even number of "a".
