SCS

SCS performs subcellular-resolution cell segmentation by integrating imaging and sequencing data from subcellular spatial transcriptomics to assign transcript-detecting spots to cells for analysis of RNA localization and tissue organization.

Key Features:

• Data integration: Integrates imaging and sequencing data from subcellular spatial transcriptomics technologies for joint analysis.
• Transformer neural network: Uses a transformer neural network to adaptively learn each spot's position relative to its cell center.
• Spot assignment: Assigns spots to cells with high precision based on learned positional relationships.
• Segmentation performance: Demonstrated higher cell segmentation accuracy, increased cell detection, and more realistic cell size estimates compared to traditional image-based methods.
• Subcellular RNA analysis: Enables subcellular analysis of RNAs by leveraging precise spot assignments to study RNA localization.
• Validation datasets: Evaluated on two novel subcellular spatial transcriptomics technologies.

Scientific Applications:

• Subcellular-resolution cell segmentation: Produce cell masks and assignments at subcellular spatial transcriptomics resolution.
• RNA localization studies: Analyze subcellular distribution of RNAs within assigned cells.
• Tissue organization and cell-cell interactions: Investigate tissue architecture and cell-cell relationships using spot-level assignments.

Methodology:

Integrates imaging and sequencing data and applies a transformer neural network that adaptively learns each spot's position relative to the cell center to assign spots to cells.