SPROUT: spectral sparsification helps restore the spatial structure at single-cell resolution.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) provides high-resolution gene expression data but lacks spatial information.
• Spatial transcriptomics captures tissue structure but often at lower cellular or gene resolution.
• Ligand-receptor interactions are crucial for understanding cell-cell communication but can be confounded by non-specific signals.

Purpose of the Study:

• To develop a computational tool, SPROUT, for reconstructing single-cell resolution spatial structures from transcriptomics data.
• To address the limitations of existing methods in accurately representing cell proximity and intercellular interactions.
• To enable the discovery of spatially resolved ligand-receptor interactions.

Main Methods:

• SPROUT utilizes transcriptomics data to infer spatial organization by diminishing pseudo ligand-receptor affinities.
• Key techniques include partial correlation, spectral graph sparsification, and spatial coordinate refinement.
• The package embeds estimated interactions into a low-dimensional space using a cross-entropy objective.

Main Results:

• SPROUT achieved high accuracy in reconstructing spatial structures, with shape Pearson correlations from 0.91 to 0.97 on mouse and human datasets.
• The tool successfully performed *de novo* reconstruction, correlating well with immunohistochemistry-informed structures (0.68 and 0.89).
• SPROUT facilitates the identification of dominant ligand-receptor pairs between neighboring cells at single-cell resolution.

Conclusions:

• SPROUT effectively reconstructs single-cell resolution spatial structures from transcriptomics data.
• The software package overcomes limitations of scRNA-seq and spatial transcriptomics by accurately modeling intercellular affinities.
• SPROUT is a valuable tool for dissecting tissue architecture and cell-cell communication in complex biological systems.