stPlus: a reference-based method for the accurate enhancement of spatial transcriptomics.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Spatial transcriptomics enables simultaneous gene expression profiling and cell mapping.
• Imaging-based spatial transcriptomics offers high resolution but faces limitations in gene number and detection sensitivity.
• Current enhancement methods for spatial transcriptomics struggle with accurate gene expression prediction and cell-population identification.

Purpose of the Study:

• To develop a novel reference-based method, stPlus, for enhancing spatial transcriptomics data.
• To improve the accuracy of gene expression prediction and cell-population identification in spatial transcriptomics.
• To provide a robust and scalable tool for analyzing diverse spatial transcriptomics datasets.

Main Methods:

• stPlus utilizes an auto-encoder with a specialized loss function for joint embedding.
• It predicts spatial gene expression using a weighted k-nearest-neighbor approach.
• A clustering-based strategy is employed to systematically assess enhancement performance.

Main Results:

• stPlus demonstrates superior performance over baseline methods, evidenced by higher gene-wise and cell-wise Spearman correlation coefficients.
• Enhanced data from stPlus facilitates improved identification of cell populations and characterization of spatial cell heterogeneity.
• The method shows robustness and scalability across datasets with varying gene detection sensitivity, sample sizes, and numbers of spatially measured genes.

Conclusions:

• stPlus effectively leverages single-cell RNA sequencing data to enhance spatial transcriptomics.
• The tool improves the accuracy of gene expression prediction and cell identification, crucial for understanding tissue architecture.
• stPlus is expected to significantly advance the analysis of spatial transcriptomics data.