Encrypting and storing sensitive data securely in DNA

Tape- and disk-based data storage degrades and can become obsolete, requiring rewriting every decade or so. Cloud- or server-based storage requires a vast amount of electricity; in 2011 Google's server farms used enough electricity to power 200,000 U.S. homes. Furthermore, old-school methods require lots and lots of space.

IBM estimated 1,000 Gb of information in book form would take up seven miles of bookshelves. In fact, Sandia recently completed a 15,000-square-foot building to store 35,000 boxes of inactive records and archival documents.

"Historically, the national laboratories and the U.S. government have a lot of highly secure information that they need to store long-term. I see this as a potentially robust way of storing classified information in the future to preserve it for multiple generations," said Bachand. "The key is how do you go from text to DNA and do that in a way that is safe and secure."

Bachand was inspired by the recording of all of Shakespeare's sonnets into 2.5 million base pairs of DNA—about half the genome of the tiny E. coli bacterium. Using this method, the group at the European Bioinformatics Institute could theoretically store 2.2 petabytes of information—200 times the printed material in the Library of Congress—in one gram of DNA.

Marlene Bachand, a biological engineer at Sandia and George Bachand's spouse, added, "We are taking advantage of a biological component, DNA, and using its unique ability to encode huge amounts of data in an extremely small volume to develop DNA constructs that can be used to transmit and store vast amounts of encrypted data for security purposes."

The Bachands' project, funded by Sandia's Laboratory Directed Research & Development program, has successfully moved from the drawing board to letterhead. Using a practically unbreakable encryption key, the team has encoded an abridged version of a historical letter written by President Harry Truman into DNA.

They then made the DNA, spotted it onto Sandia letterhead and mailed it—along with a conventional letter—around the country. After the letter's cross-country trip, the Bachands were able to extract the DNA out of the paper, amplify and sequence the DNA, and decode the message in about 24 hours at a cost of about $45.

To achieve this proof-of-principle, the first step was to develop the software to generate the encryption key and encrypt text into a DNA sequence. Andrew Gomez worked on this while he was an intern at Sandia; he is now at Senior Scientific, a nanomedicine company at the University of New Mexico's Science and Technology Park.

DNA is made up of four different bases, commonly referred to by their one letter abbreviations: A, C, G and T. Using a three-base code, exactly how living organisms store their information, 64 distinct characters—letters, spaces and punctuation—can be encoded, with room for redundancy.

For example, spaces make up on average 15 to 20% of the characters in a text document, an encryption key could specify that TAG, TAA and TGA each code for "space" while GAA and CTC could code for "E." This would reduce the amount of repetition—technically challenging for making and reading DNA—and make brute-force hacking more difficult.

The team's first test was to encode a 180-character message, about the size of a tweet. Encoding the message into 550 bases was easy; actually making the DNA was hard.

Since successfully encoding, making, reading and decoding the 180-character message and the 700-character Truman letter, the Bachands are now working on even longer test sequences. However, what the Bachands really want to do is move beyond tests and apply their technique to national security problems.

Two possible applications the team has identified are storing historical classified documents and barcoding/watermarking electromechanical components, such as computer chips made in the Microsystems and Engineering Sciences Applications complex, Sandia's Department of Defense-certified fabrication facility, prior to storage.

George Bachand imagines encoding each component's history—when it was manufactured, the lot number, starting material, even the results of reliability tests—into DNA and spotting it onto the actual chip.

Instead of having to find the serial number and look up that metadata in a digital or paper-based database, future engineers could swab the chip itself, sequence the DNA and get that information in a practically tamper-proof manner.

To test the feasibility, Marlene Bachand spotted lab equipment with a test message, and was able recover and decode the message, even after months of daily use and routine cleaning. DNA spotted onto electronic components and stored in cool, dark environments could be recoverable for hundreds of years.

Another, more straightforward application for the Bachands' DNA storage method would be for historical or rarely accessed classified documents. DNA requires much less maintenance than disk- or tape-based storage and doesn't need lots of electricity or tons of space like cloud- or paper-based storage.