Development of an Efficient File Access Technology for DNA Memory using Selective Single-Nucleotide Synthesis Mechanism

Abstract

Synthetic DNA has emerged as a promising information storage medium because of its exceptional data density, long-term stability, and compatibility with established oligonucleotide synthesis and sequencing technologies. However, the practical utility of DNA memory depends not only on data encoding and decoding, but also on the ability to efficiently access a desired file or subset from a highly complex oligonucleotide pool. Conventional access methods have mainly relied on selective hybridization, particularly polymerase chain reaction （PCR）-based amplification using subset-specific primer pairs. Although effective in many settings, such approaches require dedicated primer design for each target subset, consume a substantial fraction of oligonucleotide length for access regions, and provide limited flexibility for hierarchical or programmable file retrieval. This dissertation presents an efficient file access technology for DNA memory based on a selective single-nucleotide synthesis mechanism. The proposed method performs subset selection through cyclic nucleotide synthesis and blocking on barcode positions rather than through target-specific selective hybridization. After hybridization of a universal primer to a common priming region, nucleotides complementary to a target barcode are introduced with reversible terminators, whereas non-target bases are introduced with irreversible terminators. Through iterative coupling and deprotection cycles, only oligonucleotides carrying the desired barcode remain extendable and are subsequently recovered as selected subsets. This mechanism enables barcode recognition at single- nucleotide resolution and supports quaternary addressing, allowing 4N subsets to be theoretically encoded by an N-nucleotide barcode. In addition, because the selection process is cyclic, hierarchical access to multiple lower-level subsets sharing a common barcode prefix can be implemented in a programmable manner. The feasibility of the method was first demonstrated using a five-oligo model system with distinct barcodes and lengths, in which single, hierarchical, and multiple subset selection were successfully validated by polyacrylamide gel electrophoresis. The scalability of the method was then evaluated using a hierarchically encoded DNA memory library composed of 12,000 oligonucleotides containing four- nucleotide barcodes. In this complex library, the proposed strategy enabled efficient access to target subsets across different hierarchical levels, including file-level and folder-level retrieval. Selection efficiency was quantitatively analyzed using next-generation sequencing, showing substantial enrichment of target subsets with limited bias across distinct barcodes. The method also enabled recovery of rare subsets addressed by longer barcodes and supported targeted subset replacement through negative synthesis and selection, demonstrating the possibility of replacement-based and structurally organized information access within DNA libraries. These results establish selective single-nucleotide synthesis as a programmable molecular access mechanism for DNA memory. By reducing dependence on subset-specific primers, minimizing indexing overhead, and enabling hierarchical file retrieval, the proposed strategy expands the design space of scalable DNA memory systems and provides a useful framework for broader programmable nucleic acid applications.