clustSIGNAL

clustSIGNAL performs adaptive smoothing and clustering of spatially informed gene expression data using entropy-based neighbourhood heterogeneity measurements to weight smoothing intensity.

Key Features:

• Entropy-based heterogeneity measurement: clustSIGNAL uses entropy to measure heterogeneity of cell neighbourhoods.
• Weighted adaptive smoothing: Homogeneous neighbourhoods are smoothed more and heterogeneous neighbourhoods are smoothed less.
• Data sparsity handling: The adaptive smoothing approach overcomes data sparsity in gene expression data.
• Spatial context integration: The method incorporates spatial context into the gene expression data.
• Downstream compatibility: The resulting smoothed gene expression data is used for clustering and could be used for other downstream analyses.

Scientific Applications:

• Spatial transcriptomics clustering: clustSIGNAL clusters gene expression data while preserving spatial information from cell neighbourhoods.
• Spatial gene expression analysis: The smoothed gene expression output enables spatially informed downstream analyses beyond clustering.

Methodology:

clustSIGNAL measures neighbourhood heterogeneity using entropy, then performs weighted adaptive smoothing where smoothing intensity varies inversely with neighbourhood heterogeneity, producing spatially informed smoothed gene expression data for clustering.