Single-frame super-resolution sequencing method based on lattice prior

Area of Science:

• Genomics
• Biotechnology
• Microscopy

Background:

• High-throughput sequencing demands higher resolution for DNA Nanoball (DNB) arrays to reduce costs and increase throughput.
• Existing super-resolution methods like structured illumination microscopy (SIM) are too slow for practical high-throughput applications.

Purpose of the Study:

• To develop a fast, accurate, and cost-effective super-resolution algorithm for high-throughput sequencing.
• To overcome the speed limitations of traditional multi-frame super-resolution techniques.

Main Methods:

• Proposed SFS-seq, a non-deep-learning, single-frame super-resolution algorithm.
• Leveraged sequencing scene priors to decouple DNB brightness from wide-field images.
• Algorithm complexity is O(N) per iteration, requiring no training data or GPU.

Main Results:

• SFS-seq achieved high brightness reconstruction consistency with SIM (correlation coefficient of 0.9852).
• Demonstrated significantly improved base calling accuracy compared to direct wide-field image analysis.
• Validated the algorithm's efficiency and effectiveness in a sequencing context.

Conclusions:

• SFS-seq offers a viable single-frame super-resolution solution for high-throughput sequencing.
• The method combines speed, accuracy, and low cost, addressing key limitations in current sequencing technologies.