Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about a revolutionary approach to DNA-based data storage through this 20-minute conference presentation that addresses the exponential growth of global data projected to reach 400 zettabytes by 2028. Discover how DNA storage offers exceptional information density, enhanced durability, and negligible power consumption compared to traditional storage technologies, while exploring the practical challenges of synthesis and sequencing errors, computational complexity, and scalability issues. Examine the innovative DNAformer pipeline, a modular solution that integrates Deep Neural Networks trained on simulated data, Tensor-Product-based Error-Correcting Codes, and safety margin mechanisms to overcome DNA storage limitations. Explore the three-step retrieval process including clustering of noisy reads using predesigned indices, reconstruction through DNNs combining convolution and transformer layers with confidence scoring, and decoding via TP-based error correction codes. Understand how this unified approach achieved remarkable results on 3.1MB of data across two sequencing technologies, demonstrating a 3200× speed improvement, 40% accuracy increase, and 1.6 bits per base code rate under high-noise conditions. Gain insights into the Conditional Probability Logic algorithm that reprocesses low-confidence clusters to boost reconstruction accuracy, and learn how this breakthrough addresses critical computational bottlenecks while ensuring scalability for future data storage solutions.
