spatialRSP (Spatial Radar Scanning Plot) is a Python toolkit for quantifying spatial signal patterns in 2D embeddings of high-dimensional data. It provides interpretable, statistically rigorous metrics that reveal how signals—such as gene expression, image features, or embedding vectors—are distributed across space.
By computing two key metrics from Radar Scanning Plots (RSPs)—A1 (coverage bias) and A2 (angular skew)—spatialRSP captures spatial enrichment, dispersion, and directional asymmetry across expression thresholds. This enables robust detection of biologically or structurally meaningful patterns that are often missed by clustering or visual inspection.
Originally developed for spatial transcriptomics, spatialRSP is broadly applicable to single-cell data, imaging, NLP embeddings, and more. It offers a generalizable framework for identifying spatial biomarkers, assessing heterogeneity, and interpreting complex embeddings with precision.
We are currently working on a series of tutorials to help you get started with spatialRSP. The first tutorial is available here. More tutorials will be added soon, so stay tuned!