Inferring histology-associated gene expression gradients in spatial transcriptomic studies

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• Spatially resolved transcriptomics integrates RNA abundance with tissue architecture.
• Identifying signature genes traditionally requires preliminary data grouping, potentially missing subtle patterns.
• Complex tissues present challenges for gene expression analysis due to intricate spatial relationships.

Purpose of the Study:

• To introduce Spatial Gradient Screening, a novel algorithm for supervised detection of histology-associated gene expression patterns.
• To enable cluster-free identification of gene expression patterns, overcoming limitations of traditional methods.
• To provide a comprehensive R-based open-source framework (SPATA2) for spatial transcriptomic analysis.

Main Methods:

• Development and application of the Spatial Gradient Screening algorithm.
• Utilizing spatial transcriptomic data from injured mouse cortex.
• Incorporating single-cell deconvolution and T-cell receptor sequencing (TCR-seq) data from brain tumors.
• Comparison with standard differential gene expression analysis.

Main Results:

• Spatial Gradient Screening successfully detects histology-associated gene expression without prior data clustering.
• The algorithm offers deeper insights into the spatial architecture of transcriptomes compared to standard methods.
• Demonstrated advantages and limitations of cluster-free gene expression detection.

Conclusions:

• Spatial Gradient Screening provides a powerful, cluster-free approach for analyzing spatial transcriptomic data.
• The SPATA2 framework offers valuable tools for investigating gene expression within tissue context.
• This methodology advances the study of gene expression patterns in complex biological systems.