De-Bruijn graph partitioning for scalable and accurate DNA storage processing

DNA-based data storage offers a compelling solution for long-term, high-density archiving. In this framework, accurately reconstructing high-quality encoded sequences after sequencing is critical, as it directly impacts the design of error-correcting codes optimized for DNA storage. Furthermore, efficient and scalable processing is essential to manage the large volumes of data expected in such applications.

We introduce a novel method based on de-Bruijn graph partitioning, enabling fast and accurate processing of sequencing data regardless of the underlying sequencing technology and without requiring prior knowledge of the information encoded in the oligonucleotides. Evaluated on both synthetic and real datasets, the method achieves excellent precision and recall. It is implemented in C++ within the software ConCluD and optimized for multi-core servers. Our experiments show that a dataset of 89 million reads, corresponding to a 10 GB fasta file, can be fully processed in less than a minute on a standard 32-cores server.

The ConCluD software and the scripts to reproduce the experiments from this paper are available at https://gitlab.inria.fr/pim/org.pim.dnarxiv under the GNU AGPLv3 licence. An archival snapshot of the repository is also provided at https://doi.org/10.5281/zenodo.17160067.