hoodscanR: profiling single-cell neighborhoods in spatial transcriptomics data

Area of Science:

• Spatial biology
• Computational biology
• Bioinformatics

Background:

• Understanding complex cellular neighborhoods offers insights into tissue biology.
• Accurate neighborhood identification is crucial but challenging with existing methods.
• Current methods often use hard assignment, lacking cell-specific neighborhood profiles.

Purpose of the Study:

• To develop a novel Bioconductor package for identifying and analyzing cellular neighborhoods in spatial data.
• To provide cell-specific neighborhood profiles using a probabilistic approach.
• To enable downstream analyses for deeper insights into tissue architecture and function.

Main Methods:

• Developed hoodscanR, a Bioconductor package for spatial data analysis.
• Implemented a cell-level neighborhood probability profile representing partial cell membership across neighborhoods.
• Enabled neighborhood visualization, uncertainty assessment, clustering, and differential expression analysis.

Main Results:

• hoodscanR generates cell-level neighborhood probability profiles for spatial data.
• The package supports various downstream analyses, including visualization and differential expression.
• Applied to breast and lung cancer datasets, hoodscanR characterized mixed tissue environments and identified spatial transcriptional changes.

Conclusions:

• hoodscanR offers a probabilistic approach to cellular neighborhood analysis in spatial biology.
• The package enhances the characterization of complex tissue environments and tumor heterogeneity.
• hoodscanR facilitates the discovery of spatially regulated biological processes and therapeutic targets.