SEPAR enables spatial metagene discovery and associated molecular pattern characterization in spatial transcriptomics...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Spatially resolved transcriptomics (SRT) provides high-resolution gene expression data within its spatial context.
• Interpreting SRT data to understand complex cellular and molecular organization is challenging.
• Existing computational methods often focus on global domains, neglecting localized structures.

Purpose of the Study:

• To introduce SEPAR, an unsupervised computational framework for analyzing SRT data.
• To leverage spatial metagenes and integrate gene activity with spatial neighborhood relationships.
• To enable downstream analyses for a deeper understanding of spatial gene expression.

Main Methods:

• SEPAR utilizes spatial metagenes to analyze gene activity and spatial neighborhood relationships.
• The framework supports identifying metagene pattern-specific genes and spatially variable genes (SVGs).
• It facilitates delineation of spatial domains and refinement of expression signals.

Main Results:

• SEPAR successfully identified biologically meaningful gene ontologies and cell types linked to metagene patterns.
• The framework demonstrated higher accuracy in detecting spatially variable genes (SVGs).
• SEPAR enhanced biological signals through gene refinement and uncovered molecular interactions in multi-omics data.

Conclusions:

• SEPAR offers a novel approach to analyze SRT data, focusing on localized structures.
• The framework improves the identification of SVGs and enhances biological signal interpretation.
• SEPAR provides valuable insights into spatial molecular interactions within tissues.