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

High-throughput sequencing requires breaking the resolution limit to support higher-density DNA Nanoball (DNB) arrays, thereby reducing cost and increasing sequencing throughput. Traditional multi-frame super-resolution methods, such as structured illumination microscopy (SIM), sacrifice imaging speed, leading to reduced throughput. This paper proposes a non-deep-learning single-frame super-resolution algorithm based on sequencing scene priors, named SFS-seq. It requires no training data or GPU, directly decoupling DNB brightness from a single wide-field image with O(N) per iteration. Experiments show that SFS-seq reconstructs brightness highly consistent with SIM, achieving a correlation coefficient as high as 0.9852, and its base calling accuracy significantly outperforms direct use of wide-field images. SFS-seq combines speed, accuracy, and low cost, providing a single-frame solution for high-throughput sequencing.