Bulk2Space

Bulk2Space performs spatial deconvolution of bulk RNA-seq data to recover single-cell-resolution transcriptional profiles and spatial organization by integrating single-cell and spatial transcriptomics references with a deep learning framework.

Key Features:

• Simultaneous Spatial and Cellular Heterogeneity Disclosure: Integrates spatial information with cellular heterogeneity to assign single-cell-resolution transcriptional patterns within mixed bulk RNA-seq samples.
• Utilization of Existing References: Leverages single-cell and spatial transcriptomics datasets as references to inform deconvolution.
• Deep Learning Framework: Employs a deep learning-based algorithm to model and deconvolve bulk transcriptomic signals using reference data.
• Application in Tumor Immunology and Inflammation-Induced Tumorigenesis: Reveals spatial variance among immune cells across tumor regions and elucidates molecular and spatial heterogeneity during inflammation-induced tumorigenesis.
• Discovery of Spatial Patterns and Novel Genes: Facilitates identification of spatial expression patterns and novel genes across diverse cell types.
• Validation through Bulk Transcriptomics: Validates deconvolution outputs using bulk transcriptomic data to assess robustness.

Scientific Applications:

• Neuroscience Research: Applied to bulk transcriptomes from two mouse brain regions using Spatial-seq to reconstruct the hierarchical structure of the mouse isocortex and to annotate previously unidentified cell types in the mouse hypothalamus.
• Tumor Microenvironment Analysis: Dissects spatial and cellular heterogeneity in tumor microenvironments to reveal spatial distributions and interactions of immune cells.
• Inflammation-Induced Tumorigenesis Studies: Elucidates molecular and spatial heterogeneity associated with inflammation-driven tumorigenesis.

Methodology:

Uses a deep learning framework to integrate single-cell and spatial transcriptomics references with bulk RNA-seq data to perform spatial deconvolution that accounts for cellular composition and spatial organization, and can utilize publicly available single-cell and spatial datasets.