The rapid growth of digital information necessitates advancements in storage technologies beyond current limitations. DNA offers a promising solution due to its high storage density, longevity, and energy efficiency. Existing DNA-based storage systems, however, often lack dynamic capabilities for accessing and manipulating data, hindering their practical viability. These systems usually rely on PCR for information retrieval, a process that is slow, destructive to the original DNA, and prone to off-target binding due to primer interactions with the data payload. The researchers aim to develop a dynamic DNA storage system that addresses these limitations by enabling efficient, high-density encoding, repeatable access, and in-storage file manipulations. The system should be physically scalable to massive capacities and compatible with efficient and dense information encoding. The importance of this research lies in moving DNA-based data storage closer to practical application, reducing reliance on destructive and inefficient methods, and potentially offering a superior alternative to conventional electronic storage.

Previous research has focused on improving individual steps in DNA-based storage, including encoding, synthesis, sequencing, and decoding. However, dynamic data access within the storage database has received less attention due to the slower nature of DNA synthesis and sequencing compared to electronic methods. Existing approaches, such as PCR-based access, require removal and amplification of database portions, limiting reusability and requiring encoding strategies that trade off storage density to avoid cross-interactions. While techniques like strand displacement and single-stranded DNA toeholds have been explored for DNA computation and rewritable storage, they often present trade-offs between scalability, encoding density, and reusability. This paper builds upon existing work in synthetic and molecular biology, drawing inspiration from how cells access information in their genome, specifically focusing on the potential of PCR alternatives for efficient and repeatable data access.

The researchers developed DORIS, a system employing double-stranded DNA with single-stranded overhangs (ss-dsDNA). Each ss-dsDNA strand consists of a T7 promoter, a unique single-stranded overhang sequence acting as a file address, and a data payload region. A database comprises numerous ss-dsDNA strands, where all strands belonging to a file share the same overhang address. The authors optimized the creation of ss-dsDNA using a single-primer extension method, maximizing ss-dsDNA production by varying the ssDNA:primer ratio and PCR cycles. They then demonstrated the specific separation of individual files from a mixed pool using biotinylated oligos complementary to the overhang sequences and streptavidin-coated magnetic beads. The efficiency of separation was investigated by varying overhang lengths and temperatures, corroborating findings with thermodynamic analysis using the Oligonucleotide Properties Calculator. The study compared DORIS to PCR-based systems, highlighting DORIS's ability to prevent off-target oligo binding due to the dsDNA payload blocking unwanted interactions. Monte Carlo simulations were conducted to compare the theoretical capacity and density of DORIS with PCR-based systems, showing DORIS's significant advantages in handling large databases and dense encodings. The system's reusability was demonstrated by repeatedly accessing a specific file via in vitro transcription (IVT) using the T7 promoter, reverse transcription to cDNA, and returning the original DNA to the database. The effects of IVT duration on file retention were analyzed, and the quality of IVT was assessed using ss-dsDNAs of varying lengths. Finally, saturation mutagenesis around the T7 promoter and subsequent sequencing were used to determine design criteria for optimizing transcription efficiency, demonstrating the system's scalability to large and complex ss-dsDNA pools. Lastly, proof-of-principle experiments demonstrate in-storage file operations, including locking, unlocking, renaming, and deleting files using DNA toeholds, showcasing the dynamic capabilities of DORIS.

The study's key findings highlight the superior performance of DORIS compared to PCR-based DNA storage systems. The authors demonstrated that ss-dsDNA strands could be efficiently created using a single-primer extension method, with optimal conditions identified through optimization of ssDNA:primer ratio and PCR cycles. Highly specific separation of files from a mixed pool was achieved using biotinylated oligos and magnetic beads, with separation efficiency influenced by overhang length and temperature. DORIS was shown to significantly outperform PCR-based systems in terms of avoiding off-target binding, which is a crucial factor in maintaining data integrity and scalability. Monte Carlo simulations demonstrated a dramatic increase in the theoretical capacity and density of DORIS compared to PCR-based methods, especially for large databases and high-density encodings. Repeatable file access was achieved using IVT and reverse transcription, with approximately 50% of file A strands retained after five accesses, demonstrating the reusability of the system. Experiments indicated that adjusting the time of IVT impacts the retention rate. Through saturation mutagenesis of the T7 promoter region, the researchers identified sequence variations that optimized RNA yields, demonstrating the potential for precise control over transcriptional efficiency and revealing design principles for improving information access. Finally, the successful implementation of in-storage file operations (locking, unlocking, renaming, and deleting) using DNA toeholds further underlines DORIS's capacity for dynamic data management.

The findings demonstrate that the proposed ss-dsDNA architecture with a T7 promoter and single-stranded overhangs provides a robust and scalable platform for dynamic DNA data storage. DORIS's ability to achieve high-density encoding, repeatable access, and in-storage file manipulation significantly advances the potential of DNA-based data storage systems. The isothermal nature of DORIS operations is advantageous for maintaining DNA integrity and simplifies device design. Compared to PCR, DORIS offers several key advantages: it reduces off-target binding, enabling higher data densities; it simplifies address design, making it scalable to enormous databases; and it allows for multiple uses of the same DNA database. The identification of optimal T7 promoter sequences for enhanced transcriptional efficiency further refines the system’s design, emphasizing the system's tunability. The study contributes to advancing DNA-based data storage by offering a practical and scalable system with dynamic capabilities, which are crucial for transitioning this technology from theoretical possibilities to real-world applications.

This research presents DORIS, a novel and advanced dynamic DNA-based information storage system. DORIS offers substantial improvements in scalability, density, and dynamic operation compared to existing methods. The key innovations of ss-dsDNA, T7 promoter-based transcription, and DNA toehold-mediated file manipulation enable efficient, repeatable access, and in-storage operations. The study’s findings provide valuable insights into optimizing DNA storage system design, paving the way for future research exploring diverse in-storage operations and further miniaturization of the technology. Future work should focus on improving the efficiency and accuracy of each step of the process in increasingly large and diverse datasets, exploring new materials and enzymes to enhance reusability and minimizing error rates.

While DORIS demonstrates significant advancements, some limitations exist. The current system's file retention rate after repeated accesses is not 100%, suggesting potential for optimization. Further investigation into the impact of IVT duration and the development of strategies to minimize DNA degradation during the process are warranted. The study primarily uses in vitro experiments; the translation of DORIS to in vivo systems or fully automated high-throughput platforms needs further exploration. The current simulations, while insightful, only consider a subset of the sequence space; exploring the entire sequence space for larger datasets may reveal different relationships between capacity and density.