Points2Regions

Points2Regions identifies regions with similar mRNA compositions in imaging-based spatial transcriptomics data through rapid feature extraction and efficient clustering.

Key Features:

• Multi-scale region discovery: Identifies biologically interesting regions at different spatial scales, from cellular-scale mRNA compositions corresponding to cell types to millimeter-scale compositions corresponding to tissue-level structures.
• Rapid feature extraction: Rasterizes points representing mRNAs onto a pyramidal grid for fast processing.
• Efficient clustering: Combines hierarchical clustering and k-means clustering for region identification.
• No additional data required: Performs region discovery without relying on complementary data such as pre-segmented cells or lengthy optimization.
• Performance validation: Demonstrates performance similar to state-of-the-art methods on simulated datasets while being several times faster, and identifies regions similar to those found in other studies on real-world datasets.

Scientific Applications:

• Cell type identification: Identifies cellular-scale mRNA compositions that correspond to distinct cell types.
• Tissue structure analysis: Identifies millimeter-scale mRNA compositions that correspond to tissue-level structures.
• Exploratory analysis of imaging-based spatial transcriptomics: Enables fast and efficient exploration of spatial biomarker patterns across multiple scales without preprocessing constraints.

Methodology:

Points2Regions rasterizes points representing mRNAs onto a pyramidal grid for rapid feature extraction, then applies a combination of hierarchical clustering and k-means clustering for efficient region identification.