Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn about a representation-based architecture for DNA data storage that leverages deep learning models to achieve robust image reconstruction from noisy DNA reads. Discover how this innovative approach combines an autoencoder and U-Net network to handle representation, construction, and refinement of images stored in DNA, demonstrating successful reconstruction with insertion-deletion-substitution error rates below 6%. Explore the flexible compression capabilities through feature quantization that allows balanced trade-offs between compression ratio and image quality by adjusting representation channel numbers. Understand how multiple DNA reads can enhance image quality reconstruction and examine real-world validation through wet lab experiments that successfully reconstructed images stored across 14 plasmids. Gain insights into this competitive solution for robust and efficient DNA storage systems designed for large-scale image applications, presented by Qingyuan Fan from Southern University of Science and Technology, China at the 2025 Storage and Computing with DNA Conference.
