-blue.svg){kind=icon}
R package companion to MatriSpace for identifying, quantifying, and interpreting spatially-resolved extracellular matrix (ECM) gene expression patterns in spatial transcriptomics data.
If you use matrispace in your publications, please cite our preprint: doi 10.64898/2026.04.26.720198
The online MatriSpace app provides an interactive interface over curated datasets and user uploads. The matrispace package exposes the same workflow in R for scripted analyses, custom modeling, and figure generation.
The extracellular matrix is a spatially organized signaling environment that shapes tissue architecture, niche identity, inflammation, fibrosis, and tumor progression. Spatial transcriptomics enables the study of ECM gene expression in its native tissue context, but most general-purpose workflows are not built around matrisome organization or ECM niches. matrispace brings the MatriSpace analytical framework into R so that matrisome-centered analyses can be reproduced and extended in local pipelines.
The package follows the same three stages as the MatriSpace web application: data input, matrisome profiling, and feature analysis.
Accepts Seurat objects directly and converts SpatialExperiment objects via convert_spe_to_seurat(). prepare_object() standardizes gene symbols, computes matrisome signatures, and classifies each spot into an ECM niche.
score_matrisome()— UCell scores for the six matrisome categories and curated subfamilies (Basement Membrane, Laminins, Matricellular, Mucins, Peri-vascular ECM).annotate_ecm_niches()— ECM niche classification (e.g. Interstitial, Basement membrane), with continuous per-spot niche scores.score_lr_activity()/find_lr_enrichment()— ECM-focused ligand-receptor co-expression within and across ECM niches.
- Spatial visualization:
plot_spatial_feature(),plot_matrisome(),plot_spatial_blend(),plot_lisa(). - Local spatial association:
compute_lisa(). - Global spatial autocorrelation:
compute_morans_i(). - Differential expression:
find_spatially_corrected_ecm_markers()— within-sample, two-stage Wilcoxon screen followed by aspaMMMatérn spatial mixed model.find_ecm_de_samples(pseudobulk = TRUE)— cross-sample pseudobulk testing withDESeq2, with optional cluster and blocking adjustment.
Requires R (>= 4.1.0).
install.packages("pak")
pak::pak("Theayomideo/matrispace")library(matrispace)
seurat_obj <- readRDS("my_visium_data.rds")
seurat_obj <- prepare_object(seurat_obj)
seurat_obj <- score_matrisome(seurat_obj)
seurat_obj <- annotate_ecm_niches(seurat_obj)See the package vignettes for full differential-expression, ligand-receptor, and spatial-statistics examples.