Optimization-based decoding of Imaging Spatial Transcriptomics data

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Imaging Spatial Transcriptomics (IST) enables gene expression profiling within cellular contexts at single-molecule resolution.
• Current IST methods require high-magnification imaging, limiting throughput and broad applicability.

Purpose of the Study:

• To develop a novel algorithm for decoding IST data acquired at lower magnifications.
• To improve the throughput and performance of IST by reducing reliance on high-magnification imaging.

Main Methods:

• Developed the Joint Sparse method for Imaging Transcriptomics (JSIT) algorithm.
• Incorporated codebook knowledge and sparsity assumptions into an optimization framework.
• Applied JSIT to Multiplexed Error-Robust Fluorescence in situ Hybridization data from mouse brain tissue.

Main Results:

• JSIT enables decoding of IST data at lower magnifications than standard methods.
• The algorithm demonstrates improved throughput and recovery performance compared to existing pipelines.
• JSIT is less sensitive to optical signal separation compared to current approaches.

Conclusions:

• The Joint Sparse method for Imaging Transcriptomics offers a more efficient approach to gene expression analysis in situ.
• This advancement has the potential to increase the impact and accessibility of spatial transcriptomics techniques.