DNA has been the storage medium used by nature to capture biological data for several billion years. The notion of storing digital information using synthetically-created DNA molecules has been around for decades. Over the past decade, a small number of research labs have worked to develop this technology, and have made some progress in capturing and retrieving greater quantities of information using this medium. However, stored data sets have been small, and chemically producing synthetic DNA molecules has remained slow and expensive. While the read side of the process, achieved through DNA sequencing techniques, has improved tremendously over the past few decades, writing has remained elusive. DNA as a commercial data storage technology has appeared to be far off. The promise, however, is huge, as DNA from a volumetric perspective is a million times more information-dense than today’s magnetic or flash media.

The Boston-based startup Catalog has achieved a significant leap in this technology, by taking a combinatorial, information theory-centric approach to creating synthetic DNA molecules that capture digital information. Instead of slow and expensive techniques to create custom DNA molecules a base pair at a time, we approach involves utilizing pre-fabricated, easily and inexpensively-replicable shorter DNA sequences that can be combined enzymatically to represent binary data objects. This has allowed us to leapfrog other efforts in terms of write speed by many orders of magnitude, bringing us significantly closer to being able to commercialize data storage systems using this medium. As a medium, DNA is extremely attractive not only because of its potential density – an exabyte of data in the volume of a sugar cube – but also because data sets can be reproduced quickly and inexpensively, and resulting samples can potentially remain viable for thousands of years. We are also developing approaches to perform direct processing of data stored in DNA, also using enzymatic and other biocomputing techniques. Because DNA data samples can contain many thousands of copies of a data set, there is the promise for certain types of parallelizable processing tasks to realize a major leap in performance using these techniques. This presentation will offer a primer into DNA data storage technology, describe our own unique approaches in encoding scheme, processes, and systems engineering that are leading to the fast-tracking of commercial applications around this technology, where the roadmap is headed, and what problems in digital media these developments will help to solve.