SOMDE

SOMDE identifies spatially variable genes (SVgenes) in large-scale spatial transcriptomic sequencing datasets to detect spatial patterns of gene expression across tissue microenvironments.

Key Features:

• Scalability: Optimized for large-scale datasets and reported to be approximately 5 to 50 times faster than other methods while maintaining comparable accuracy.
• Self-Organizing Map (SOM): Employs a Self-Organizing Map to cluster neighboring cells into nodes, reducing spatial data complexity by grouping similar expression profiles.
• Gaussian Process: Uses a Gaussian Process to model node-level spatial gene expression patterns for identifying spatial variability.
• Adjustable Resolution: Supports adjustable resolution for analysis, enabling results in approximately 5 minutes for datasets with over 20,000 sequencing sites.

Scientific Applications:

• Spatial gene expression analysis: Identifies SVgenes to map spatial expression patterns within tissues.
• Tissue microenvironment and cellular interactions: Enables analysis of cellular behavior and interactions within tissue microenvironments.
• Developmental biology: Supports studies of spatial gene regulation during tissue development.
• Disease pathology: Facilitates identification of spatial expression patterns relevant to disease mechanisms.
• Regenerative medicine: Aids investigation of spatial cues relevant to tissue regeneration and repair.

Methodology:

SOMDE applies a Self-Organizing Map to cluster neighboring cells into nodes, aggregates expression at the node level, and fits a Gaussian Process to model spatial gene expression, with adjustable resolution for computational efficiency.