Abstract Data storage using large DNA fragments enables low‐cost in vivo replication and offers a promising strategy for distributed data applications. However, data readout from massive, unordered sequencing reads requires alignment based on overlapping regions and is complicated by diverse sequencing errors, especially insertion and deletion (indel) errors. Here, we propose a fast and reliable readout framework in a bootstrap manner tailored for data storage using watermarked large DNA fragments. Our scheme transforms the de novo readout into a resequencing‐like workflow through multiple‐fold hidden references, substantially reducing readout complexity. The framework is compatible with sequencing platforms exhibiting diverse error profiles. For technologies with low indel rates, we employ correlation‐based identification and bit‐wise consensus to enable rapid decoding. For indel‐prone platforms, we incorporate progressive read alignment using multiple‐fold hidden references and the forward‐backward algorithm to ensure robust recovery. In vivo experiments on large DNA fragments with different coding efficiencies validated the proposed framework. Error‐free recovery was achieved using Illumina reads (raw error rate of ~0.2%) at a coverage of 0.6–2.5× and using nanopore reads (raw error rate ~5%) at a coverage of 1.6× or 4.3×. These results demonstrate the practicality and scalability of large‐fragment DNA storage for real‐world applications.
Chen et al. (Sun,) studied this question.