In a scientific first, Harvard University researches successfully transformed a 53,426-word book into DNA, the same substance that provides the genetic template for all living things. The achievement could eventually lead to the mass adoption of DNA as a long-term storage medium.
Published Thursday in the journal Science, the experiment aimed to demonstrate the viability of storing large amounts of data on DNA molecules. Since the data is recorded on individual nucleobase pairs in the DNA strand (those adenine-guanine/cytosine-thymine pairs you may be straining to remember from high school biology), DNA can actually store more information per cubic millimeter than flash memory or even some experimental storage techs, IEEE Spectrum reports.
After that came the heavy lifting: synthesizing the DNA strand, which would be 5.27 million bases long. They made the journey by splitting it into baby steps, each 96 bases long. When they were done, the book was a tiny speck of synthesized DNA that had about one-millionth the weight of a grain of sand. That’s got to look pretty attractive to anyone with a Big Data problem.
Reading the DNA book was a little easier, since all that was needed was commercially available DNA-sequencing tech. After arranging the sequence, it was easy to decode it back to binary code, and then the complete book as an HTML file. The researches said the errors introduced by the entire process were minimal, just 10 bits out of 5.27 million total.
Besides the storage density, DNA storage has two more advantages. The first is longevity; DNA lasts for thousands of years (or even millions, if it’s trapped in amber). The second is future-proofing: Since DNA is the basis of all life, future societies will always have technologies available to read it (that assumes artificial intelligence doesn’t exterminate or replace human society, of course).
There are downsides, though. There’s the cost — DNA-sequencing equipment is still relatively expensive — but the data is also unchangeable once it’s encoded. DNA is strictly a write-once medium.