BIG data could be about to get much smaller. Catalog, a start-up based at the Harvard Life Lab, has announced plans for the first commercial DNA data storage service. The company says it has developed a way to cheaply store a terabyte of data – the equivalent of 40 Blu-rays – in a DNA pellet.
Using DNA as a data storage medium has long held appeal thanks to its durability and density. If kept cool and dry, DNA can reliably last for hundreds of years, so a vast data centre could be replaced by an ordinary refrigerator. But current technology makes this prohibitively expensive. For example, Twist Biosciences of San Francisco can create DNA strands that hold a , enough for a handful of mp3s, at a cost of $100,000.
Catalog’s chief science officer, Devin Leake, says the firm can do much better. From 2019, the firm says it will offer a terabyte worth of DNA-encoded data in a gram-sized pellet.
Advertisement
“More significantly, compared to other DNA synthesis approaches, our cost is five orders of magnitude lower,” says Leake. But the company will need to bring costs down much further to compete with the price of cloud-storage on tape, which Amazon currently sells for .
Co-founder Hyunjun Park says that Catalog’s price hinges on a new approach. Rather than encoding a string of data onto a single DNA molecule, it produces a small core “alphabet” of DNA sequences that correspond to snippets of binary data and can be pooled together into much larger files.
Think of it this way: instead of writing every single word of a book by hand, the Catalog method uses the pre-made strands like the typeface letters in a printing press. Synthesising 13 sets of nine molecules, for example, would permit 913 different possibilities, or more than 2 trillion data sequences.
“The incremental costs of printing a different ‘book’ are much lower, because we already have the ‘typefaces’,” says Park. “We just have to arrange them differently.”
“From 2019, Catalog says it will offer a terabyte worth of DNA-encoded data in a gram-sized pellet”
Park says Catalog is currently building a machine to do this arrangement efficiently, but wouldn’t reveal exactly how it works. It is based on a standard piece of equipment called a liquid handler, used to automatically place liquids in a designated container, but is designed to perform many more parallel operations.
Reinhard Heckel, who studies DNA storage at Rice University in Houston, Texas, says it is difficult to know whether such dramatic improvements in cost and speed are possible in the near future until Catalog publishes its research or reveals what its technology is based on.
“The firm would need to generate DNA molecules many orders of magnitude faster and cheaper than what people have been able to do in recently published research,” he says. “I’m curious to see what its technology is and whether it can really, as promised, be scaled up that significantly.”
“It is really exciting to see entrepreneurial activity in this space,” says Luis Ceze at the University of Washington. Catalog’s use of a “library” of pre-made strands has the potential to achieve cost savings, he says, but needs a sophisticated fluid-handling machine to manipulate the many strands required. “I look forward to hearing more about Catalog’s magic.”
In addition to the high costs of writing, DNA storage does have another downside: it is slow to read back the data once it has been encoded. In February, Ceze and his colleagues achieved new heights by showing they could access random selections from within 200 megabytes of DNA-encoded data without having to read all 13 million nucleotides that made up the data trove.
This limitation means Catalog’s first customers are likely to be companies or government agencies that need long-term storage, rather than instant-access to files, says Park. This could include confidential records or video footage that need to be kept on-site, rather than on distant cloud servers.
Park also notes that IARPA, the research arm of the US intelligence community, has issued a , including DNA, which could indicate potential future customers.
This article appeared in print under the headline “The world’s smallest hard drive”
