scfeatureprofiler 1.0.3

A powerful, multi-interface Python package for deep characterization of single-cell feature expression patterns.

Single-Cell Feature Profiler

A powerful, fast, and multi-interface Python package for deep characterization of single-cell feature expression patterns.

scfeatureprofiler provides a suite of statistical tools to analyze single-cell data (e.g., scRNA-seq, CITE-seq) and answer two fundamental questions:

1. Feature Activity: In which cell groups is a feature actively expressed?
2. Marker Discovery: Which features are specific markers for each cell group?

The package is designed for performance, with a parallelized backend that can handle extremely large datasets, including out-of-core analysis for data that doesn't fit into memory.

Key Features

• Multi-Interface: Use it as a Python library in your Jupyter notebooks or as a command-line tool for script-based workflows.
• Flexible Input: Works directly with AnnData objects, pandas.DataFrame, or numpy arrays.
• Comprehensive Statistics: Calculates normalized expression scores, percentage of expressing cells, specificity scores (Tau and Gini), and robust statistical significance (FDR).
• High Performance: Parallelized using joblib to use all available CPU cores for rapid analysis.
• Scalable: Natively supports out-of-core computation for AnnData objects stored on disk, enabling analysis of millions of cells.
• Lightweight: Minimal dependencies, making it easy to integrate into existing analysis environments.

Installation

You can install scfeatureprofiler directly from PyPI:

pip install scfeatureprofiler

To include support for AnnData objects, install with the [anndata] extra:

pip install scfeatureprofiler[anndata]

To install all dependencies for development, use:

# Clone the repository first
git clone https://github.com/zqzneptune/SingleCellFeatureProfiler.git
cd SingleCellFeatureProfiler
pip install -e ".[all]"

Quick Start

scfeatureprofiler is designed to be intuitive. Here are two examples for the most common use cases.

1. Python API: Find Marker Genes for Clusters

This is the most common use case inside a Jupyter notebook after you have performed clustering.

import scanpy as sc
from scfeatureprofiler import find_marker_features

# 1. Load your clustered single-cell data
#    (This example assumes you have an AnnData object)
adata = sc.read_h5ad("path/to/your_clustered_data.h5ad")

# 2. Find marker features for your clusters
#    'leiden' is the column in adata.obs containing cluster labels.
marker_dict = find_marker_features(
    data=adata,
    group_by='leiden'
)

# 3. Print the results
for cluster, markers in marker_dict.items():
    print(f"Cluster {cluster} Markers: {', '.join(markers[:10])}...")

# 4. (Optional) Use the results directly with Scanpy for plotting
import scanpy as sc
sc.pl.dotplot(adata, marker_dict, groupby='leiden')

2. Command-Line (CLI): Get Full Profiles for a Gene List

If you have a CSV file of expression data and want to get a detailed statistical report for a few genes of interest without writing a script.

Input Files:

• expression.csv: A cells-by-genes matrix.
• cell_groups.csv: A file mapping cell IDs to group labels.

Command:

scprofiler profile \
    --input example/expression.csv \
    --group-by example/cell_groups.csv \
    --features "CD4,CD8A,GNLY,MS4A1" \
    --output gene_profiles.csv

Output (gene_profiles.csv): This will produce a detailed CSV file with statistics for each gene in each cell group, ready for analysis in Excel or another program.

feature_id,group,norm_score,pct_expressing,fdr_presence,fdr_marker,...
CD4,T-cell Helper,0.98,85.4,1.2e-50,3.4e-30,...
CD4,B-cell,0.05,2.1,0.89,1.0,...
CD8A,T-cell Cytoxic,0.99,92.1,4.5e-60,8.1e-45,...
...

Available CLI Commands

• scprofiler profile: Generate a full statistical profile for features.
• scprofiler activity: Identify in which groups a list of features are "ON".

Use scprofiler --help or scprofiler profile --help for a full list of options.