Accurate imputation of pathway-specific gene expression in spatial transcriptomics with PASTA

Area of Science:

• Genomics and Computational Biology
• Spatial Transcriptomics
• Disease Oncogenesis Research

Background:

• Mapping the transcriptome at single-cell resolution within its spatial context is crucial for understanding disease mechanisms.
• Current in-situ technologies offer high spatial resolution but are limited in the number of genes measured.
• Computational methods exist to predict missing gene expression in spatial transcriptomics using scRNA-seq, but often yield variable results.

Purpose of the Study:

• To develop a novel computational method for imputing pathway gene expression in spatial transcriptomics data.
• To improve the accuracy, robustness, and biological relevance of gene expression predictions in spatial datasets.
• To overcome the limitations of current methods by integrating pathway information, cell type, and spatial proximity.

Main Methods:

• Introduction of PASTA (PAthway-oriented Spatial gene impuTAtion), a new spatial pathway expression imputation method.
• Leveraging cell type identity and spatial proximity of cells to enhance prediction accuracy.
• Integrating pathway information into the imputation process to improve robustness and biological relevance.

Main Results:

• PASTA demonstrates superior performance compared to existing methods on both simulated and real-world spatial transcriptomics datasets.
• The method successfully imputes pathway gene expression with improved stability.
• PASTA enhances the biological significance of predictions in spatial transcriptomics analysis.

Conclusions:

• PASTA offers a significant advancement in spatial pathway expression imputation for spatial transcriptomics.
• The method's ability to leverage cell type and spatial context improves prediction quality and biological interpretability.
• PASTA is a valuable tool for investigating disease oncogenesis and progression using high-resolution spatial data.