Nanopore decoding with speed and versatility for data storage.

Area of Science:

• Biotechnology
• Bioinformatics
• Data Storage

Background:

• Nanopore sequencing technology offers high throughput and accuracy for DNA data storage.
• Existing decoding designs for nanopore sequencing have inefficiencies in read density, error rates, or compute time.
• There is a need for flexible and efficient decoding designs tailored to nanopore capabilities for DNA data storage.

Purpose of the Study:

• To design a novel single-read per-strand decoder for DNA data storage using nanopore technology.
• To improve throughput, reduce error rates, and enable efficient reading of both DNA and RNA molecules.
• To develop a soft decoding algorithm suitable for parallel processing on GPUs.

Main Methods:

• Developed a new single-read per-strand decoder utilizing a novel soft decoding algorithm.
• Implemented and evaluated hard and soft decoders, including a new Alignment Matrix Trellis (AMT) soft decoder, on HEDGES convolutional code.
• Utilized GPU parallelization for the soft decoding algorithm to enhance computational speed.

Main Results:

• The new AMT soft decoder achieved a 257x throughput improvement over prior state-of-the-art soft decoders, with a byte error rate of 3.52%.
• Achieved read densities of 0.33 bits/base, a 4x increase compared to previous MSA-based decoders.
• RNA molecules showed 85% of the error-free read rate compared to DNA molecules.

Conclusions:

• The developed soft decoder significantly enhances the efficiency and scalability of nanopore-based DNA data storage.
• The GPU-parallelizable algorithm allows for broader exploration of system designs and higher read densities.
• Nanopore sequencing is a viable and increasingly efficient technology for DNA data storage, applicable to both DNA and RNA molecules.