updated 08:10 pm EDT, Mon August 20, 2012
Practical application still years away
Harvard geneticist and bioengineers George Church and Sriram Kosuri with the assistance of John Hopkins Yuan Gao have stored 5.5 petabits (700 terabytes) of data in a single gram of DNA. This accomplishment, published in Science, bests the previous DNA data storage record by more than 1000 times.
Researchers have been evaluating DNA as a potential storage medium for some time. The very nature of genetic material makes it attractive for storage: it is dense, requires no moving parts when stored, and durable. Instead of data being stored with magnetic media, short DNA strands storing 96 bits, with each of the base pairs representing a binary number. In this case, Adenosine and Thymine represent zero, and the Guanine and Cytosine pair represent one. In much the same way a TCP/IP packet arrives across a network or the internet, each strand of DNA is encoded with a 19-bit address block, allowing a large quantity of DNA stored volumetrically to be decoded and sorted so usable data can be extracted, 96 bits at a time.
The original human genome project, with 3 billion base pairs) took from 1990 to 2003 to completely map out. More modern laboratory equipment with advanced computational resources can accomplish the same task much more quickly, but still on the order of hours to "brew" the DNA and decode it, rather than minutes or seconds which would be required for random-access memory for computers or on-demand data storage. At this time, the data storage envisioned by Church, Kosuri, and Gao is not rewritable.
The researchers envision biological-based storage as long term archival storage, recording with multiple redundant backups, everything possible. For comparison, the same amount of data stored in the one gram of DNA would require about 170 of the recently-announced 4TB hard drives.