analysis

Analysis

Spatial statistics and analysis algorithms for spatial transcriptomics data, providing autocorrelation testing, spatially-aware clustering, cell type deconvolution, and neighborhood enrichment analysis.

Contents

File	Purpose
__init__.py	Exports autocorrelation, clustering, deconvolution, neighborhood
autocorrelation.py	Moran's I, Geary's C, LISA, Getis-Ord G*, variograms
clustering.py	Spatial graph construction, Leiden/Louvain clustering, BayesSpace-style domains
deconvolution.py	NNLS and NMF-based cell type deconvolution for multi-cell spots
neighborhood.py	Neighborhood enrichment, co-localization, Ripley's K, niche detection

Key Functions

Function	Description
autocorrelation.spatial_weights_matrix()	Build spatial weights matrix (KNN, distance-band)
autocorrelation.morans_i()	Global Moran's I spatial autocorrelation test
autocorrelation.local_morans_i()	Local Moran's I (LISA) for hotspot detection
autocorrelation.getis_ord_g()	Getis-Ord G* statistic for clustering
clustering.build_spatial_graph()	Build KNN/Delaunay/radius spatial graph
clustering.spatial_cluster()	Spatially-aware clustering combining expression and location
clustering.spatial_domains()	BayesSpace-style spatial domain identification
deconvolution.deconvolve_spots()	Deconvolve spot expression into cell type proportions
neighborhood.neighborhood_enrichment()	Compute cell type co-localization enrichment
neighborhood.ripley_k()	Ripley's K function for spatial point patterns
neighborhood.niche_detection()	Identify spatial niches from cell type composition

Usage

from metainformant.spatial.analysis import autocorrelation, clustering, neighborhood

W = autocorrelation.spatial_weights_matrix(coordinates, method="knn", k=6)
result = autocorrelation.morans_i(expression_values, W)
clusters = clustering.spatial_cluster(expression, coordinates, n_clusters=8)
enrichment = neighborhood.neighborhood_enrichment(coordinates, cell_types)