Archival data storage using DNA proximity graphs

Abstract

DNA offers a promising medium for long-term archival data storage, but the high cost of synthesizing data-specific sequences remains a major barrier to its adoption. Here we introduce SPAN (Synthesis-free Proximity-graph Archiving in Nucleic acids), a synthesis-independent storage architecture in which digital information is written as a physical pattern and encoded in the spatial relationships of randomly barcoded DNA molecules. Barcodes are deposited onto a patterned surface, and an enzymatic reaction records pairs of neighboring molecules. Sequencing these pairs produces a proximity graph from which the original information can be reconstructed. We demonstrate this principle experimentally by writing a 136-bit pattern using photolithography and correctly recovering 114 bits. Simulations further establish the physical parameters governing reconstruction accuracy. Our cost analysis shows that the demonstrated system already has lower estimated writing costs than existing DNA synthesis-based methods, and projects a further five-order-of-magnitude reduction using established optical writing technologies. Although SPAN requires more expensive readout, this trade-off makes it a promising candidate for deep archival storage, where information is written once and rarely, if ever, retrieved.