pyNebulosa

Single-cell data visualization using kernel gene-weighted density estimation.

pyNebulosa is a Python package (ported from the original R/Bioconductor package) that recovers gene expression signals lost to dropout in single-cell data by using weighted kernel density estimation on low-dimensional embeddings. It integrates natively with scanpy and AnnData.

Motivation

Standard feature plots in single-cell analysis are plagued by sparsity — many cells show zero expression even for marker genes, making it hard to see biological signal. pyNebulosa addresses this by computing a gene-weighted kernel density estimate (KDE) on the embedding space, effectively smoothing expression over neighbouring cells.

R vs Python

pyNebulosa was originally an R/Bioconductor package (Nebulosa) for Seurat and SingleCellExperiment objects. This Python port provides an equivalent API for the scanpy/AnnData ecosystem, producing identical density estimates.

library(Nebulosa)
library(Seurat)
pbmc <- pbmc3k.SeuratData::pbmc3k.final
plot_density(pbmc, "NKG7")
plot_density(pbmc, c("NKG7", "GNLY"),
             joint = TRUE)

import scanpy as sc
import pynebulosa as nb
adata = sc.datasets.pbmc3k_processed()
nb.plot_density(adata, "NKG7")
nb.plot_density(adata, ["NKG7", "GNLY"],
                joint=True)

Installation

pip install pynebulosa

With scanpy (recommended):

pip install "pynebulosa[scanpy]"

Quick start

import scanpy as sc
import pynebulosa as nb

adata = sc.datasets.pbmc3k_processed()

# Single gene density plot
nb.plot_density(adata, "NKG7")

Multi-marker panels

Visualize multiple markers at once to identify cell populations:

nb.plot_density(adata, ["MS4A1", "NKG7", "CST3", "PPBP"], ncols=4)

Joint density

Find co-expressing cell populations by computing the product of individual gene densities:

nb.plot_density(adata, ["NKG7", "GNLY"], joint=True)

API

Key parameters

How it differs from similar tools

Documentation

• Full documentation (ReadTheDocs)
• Tutorial notebook (PBMC 3K walkthrough)

Contributing

We welcome contributions of all kinds! If you'd like to contribute to pyNebulosa, please:

1. Fork the repository and create a feature branch.
2. Make your changes and add tests where appropriate.
3. Ensure all tests pass with pytest tests/ -v.
4. Submit a pull request describing your changes.

Please report bugs and feature requests on the issue tracker.

License

pyNebulosa is licensed under the MIT License.

Citation

If you use pyNebulosa, please cite:

Alquicira-Hernandez, J., Powell, J.E. Nebulosa recovers single-cell gene expression signals by kernel density estimation. Bioinformatics, 37(16), 2485-2487, 2021. doi:10.1093/bioinformatics/btab003