VISTA Uncovers Missing Gene Expression and Spatial-induced Information for Spatial Transcriptomic Data Analysis.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Understanding cell activities in spatial context is crucial for deciphering tissue function.
• Single-cell RNA sequencing (scRNA-seq) provides cell-level gene expression but lacks spatial information.
• Subcellular spatial transcriptomics (SST) offers high-resolution spatial gene expression but profiles a limited gene set.

Purpose of the Study:

• To introduce VISTA, a novel computational approach for predicting unobserved gene expression in spatial transcriptomics data.
• To enhance the utility of SST data by enabling comprehensive gene profiling.
• To facilitate deeper insights into spatially resolved cellular states and functions.

Main Methods:

• VISTA employs variational inference and geometric deep learning to jointly model scRNA-seq and SST data.
• The method incorporates uncertainty quantification for robust predictions.
• The approach is designed for efficient analysis of large-scale spatial transcriptomics datasets.

Main Results:

• VISTA demonstrates superior performance in gene expression imputation compared to existing methods.
• The approach shows satisfactory time efficiency and memory consumption for large datasets.
• Imputed gene expression data enables various downstream analyses.

Conclusions:

• VISTA effectively predicts gene expression in spatial transcriptomics data, bridging the gap between scRNA-seq and SST.
• The method unlocks new possibilities for spatial transcriptomics analysis, including identifying spatially variable genes and ligand-receptor interactions.
• VISTA advances the study of spatial biology by providing a more comprehensive view of gene activity within tissues.