In flash memory, as with everything else in tech these days, making things smaller than about 20 nanometers is very, very difficult. Building transistors at 14 nm, then 10nm, allows engineers to pack more and more information into the same physical space — a process that has taken us from 64-megabyte solid state drives to 32-gigabyte keychains. Yet there is one data storage medium that works in units many times smaller than even the most advanced computers, one that’s been in use in one form or another for quite some time indeed:DNA. A team of scientists presenting at the 250th National Meeting & Exposition of the American Chemical Society say they can use genetic material to safely store information — for 2000 years, or more.
DNA is the data storage molecule of biology. In the case of the human genome, it packs about 750 megabytes of raw information into a cell nucleus just a few micrometers across — even the best digital storage media require a few square centimeters to achieve the same thing. A DNA nucleic acid (an A, C, T, or G) is just a few angstrom across, and naturally forms a helix just a couple of nanometers in diameter. And since it works with four possible bit states, it uses a much more efficient quaternary coding system, rather than binary ones and zeroes. In other words, DNA has the capacity to pack a lot of information into a very small physical space.
Of course, DNA also makes a lot of copying errors — cancer rates can speak to that, all on their own. It’s important, when storing data, to store it accurately, so scientists must figure out a way to shore up many of DNA’s inherent fidelity problems. Other teams have used DNA for data storage, but this team focused on keeping that data intact for as long as possible. They put their data-loaded DNA fragments inside silica spheres, giving them a measure of protection, then heated the samples for a week at 160°F (70°C). They argue that this is equivalent to about 2,000 years of storage at 50°F, or 10°C.
As an aside, here’s a fun little exercise that illustrates the size of a drop of water: how many molecules of water are there in a drop? Well, there are about 20 drops in a mL, and a mL of water weighs a gram, and a gram of water is 1/18th of a mole, and a mole of water has 6.022*1023 molecules… so there’s (6E23/(18*20)) = 1.7*1021, or 1,700,000,000,000,000,000,000 molecules. The point is that, seen from the atomic perspective, a droplet of water is really big.
Now, imagine how much information could be packed into that same drop of water, with a storage medium as efficient as DNA. At that point, the barrier becomes not how much data you can physically fit, but how safe and reliable you want it to be. One of the best ways to make sure you accurately preserve DNA data is to store multiple copies together, so you can later compare multiple versions of the final “file” to determine which, if any, has acquired damage during storage. Combined with the physical size of the protective silica spheres, this does reduce the total possible data that can be stored per volume of liquid.
Another big downside to DNA storage is how slow and clunky the data retrieval process is; even with modern, high-throughput sequencing technology, reading a molecule of DNA takes orders of magnitude longer than reading a computer file; you wouldn’t want to store anything you need quickly or often, with DNA.
So what would you want to store with DNA? Anything that needs to be stored, with accuracy, for long periods of time. Time capsule data, perhaps a snapshot of Wikipedia loaded into a vial and buried for future generations, could be well suited, and since human beings are unlikely to make their own genomes obsolete any time soon, we can be sure future people retrieving this data will have some sort of DNA-reading technology in place. You can’t assume that sort of continued easy technological accessibility, on the scale of millennia, for anything else except perhaps physical writing.
Another possible aspect of DNA storage is data security. If you have information you want to keep safe, and that you’re sure you yourself won’t suddenly need quickly (say, scientific information about how to engineer a particularly deadly virus), then pioneering a uniquely difficult-to-sequence form of DNA that requires special techniques could safely effectively preserve that knowledge. Even if terrorists or some other boogeyman did manage to get their hands on the physical sample, sequencing it to read the contents would require biochemists, and lots of experimental trial and error. Cryptographic attacks wouldn’t help in the slightest — though you’d still need them if the data had been encrypted before it was stored in the DNA.
In the end, this isn’t a breakthrough that will revolutionize anyone’s day-to-day existence, but it is an intriguing step forward, and one that excites the imagination.