DEEPsc

DEEPsc imputes spatial coordinates onto scRNA-seq datasets using a spatial reference atlas to restore tissue spatial context for analyses of cell fate decisions and microenvironmental interactions.

Key Features:

• System-Adaptive Deep Learning Approach: A deep learning framework that adapts to the specific characteristics of input data to map spatial coordinates from a reference atlas onto scRNA-seq datasets.
• Comprehensive Evaluation Metrics: A set of metrics designed to evaluate accuracy, precision, and robustness of spatial mapping methods.
• Comparative Performance Across Biological Systems: Comparative analyses across four different biological systems showing comparable accuracy to existing methods with an improved balance between precision and robustness.
• Data-Adaptive Integration with Spatial Imaging: Connects scRNA-seq datasets with spatial imaging data to enable spatially informed exploration of cell fate decisions.

Scientific Applications:

• Cell Fate Analysis: Restores spatial context in scRNA-seq data to analyze cell fate decisions within native tissue architecture.
• Developmental Biology: Enables study of spatial arrangements and their influence on cell differentiation during development.
• Oncology: Supports examination of tumor microenvironment and spatial cellular interactions in cancer research.
• Tissue Engineering: Assists assessment of spatial organization and cellular context in engineered tissues.

Methodology:

Employs a system-adaptive deep learning framework to impute spatial coordinates from a spatial reference atlas onto scRNA-seq datasets, evaluates mapping using metrics for accuracy, precision, and robustness, performs comparative analyses across four biological systems, and integrates scRNA-seq with spatial imaging data.