py-SpotSweeper

Spatially-aware quality control for spatial transcriptomics.

Python port of R/Bioconductor package SpotSweeper v1.5.0.

Install

pip install spotsweeper

Quick start

import anndata as ad
from spotsweeper import local_variance, local_outliers, find_artifacts, flag_visium_outliers

# Load your spatial data as AnnData with adata.obsm['spatial']
adata = ad.read_h5ad("your_data.h5ad")

# 1. Compute local variance of mitochondrial percentage
adata = local_variance(adata, metric="subsets_Mito_percent", n_neighbors=36)

# 2. Detect local outliers in library size
adata = local_outliers(adata, metric="sum", direction="lower", log=True)

# 3. Flag systematic Visium outliers
adata = flag_visium_outliers(adata)

# 4. Find artifacts (single sample only)
adata = find_artifacts(adata, mito_percent="subsets_Mito_percent", n_order=5)

Functions

Speed vs R

Parity

All functions pass parity gates against R reference outputs:

Requirements

• Python >= 3.9
• numpy, scipy, pandas, anndata, scikit-learn, matplotlib

License

MIT

References

Totty et al. (2025) "SpotSweeper: spatially-aware quality control for spatial transcriptomics." Bioconductor.