pyturbseq

Author: Aidan Winters Email: aidanw@arcinstitute.org

Description

pyturbseq is a comprehensive Python package for processing and analysis of single-cell perturbation data, with a particular focus on CRISPR-based perturbation screens. The package provides tools for:

• Perturbation calling and quality control: Identify and validate perturbations in single cells
• Differential expression analysis: Compare gene expression between perturbed and control cells
• Interaction analysis: Detect and quantify genetic interactions in dual perturbation experiments
• Visualization: Generate publication-ready plots for perturbation screen analysis
• Data processing: Handle large-scale single-cell datasets with efficient algorithms

Key Features

• Multi-perturbation support: Handle single and dual perturbation experiments
• Flexible data input: Compatible with standard single-cell formats (AnnData, h5ad)
• Statistical analysis: Robust statistical methods for differential expression and interaction detection
• Scalable processing: Efficient algorithms for large datasets
• Comprehensive visualization: Rich plotting functions for data exploration and publication

Installation

Requirements

• Python 3.9 or higher
• Tested extensively with Python 3.9-3.13

Install from PyPI (recommended)

pip install pyturbseq

Install from source

git clone https://github.com/aidanwinters/pyturbseq.git
cd pyturbseq
pip install -e .

Development installation

For development with testing and linting tools:

pip install -e ".[dev,test]"

Quick Start

import pyturbseq as prtb
import scanpy as sc

# Load your single-cell perturbation data
adata = sc.read_h5ad("your_perturbation_data.h5ad")

# Generate perturbation matrix
prtb.utils.get_perturbation_matrix(adata, perturbation_col='feature_call')

# Calculate target gene changes
prtb.utils.calculate_target_change(adata, perturbation_column='feature_call')

# Perform differential expression analysis
# Single comparison: compare specific perturbation vs control
deg_results = prtb.de.get_degs(adata, design_col='feature_call', control_value='NTC')

# Multiple comparisons: test all perturbations vs control
all_deg_results = prtb.de.get_all_degs(adata, design_col='feature_call', control_value='NTC')

# Analyze genetic interactions (for dual perturbation data)
# Single interaction: analyze specific dual perturbation
result, prediction = prtb.interaction.norman_model(data, 'GENE1|GENE2')

# Multiple interactions: analyze specific dual perturbations
interaction_results = prtb.interaction.norman_model(data, dual_perturbation_list)

# Auto-detect and analyze all dual perturbations (default behavior)
interaction_results = prtb.interaction.norman_model(data)

# Parallel processing for large datasets
interaction_results = prtb.interaction.norman_model(data, parallel=True, processes=8)

# Generate visualizations
prtb.plot.target_change_heatmap(adata, perturbation_column='feature_call')

Main Modules

• utils: Core utilities for data processing and perturbation matrix generation
• de: Differential expression analysis tools
• interaction: Genetic interaction analysis for dual perturbation experiments
• plot: Visualization functions for perturbation screen data
• calling: Perturbation calling and quality control
• cellranger: Integration with Cell Ranger outputs

Documentation

For detailed documentation and tutorials, see the included Tutorial.ipynb notebook which demonstrates:

• Data loading and preprocessing
• Perturbation calling
• Differential expression analysis
• Interaction analysis
• Visualization workflows

Citation

If you use pyturbseq in your research, please cite:

Winters, A. (2024). pyturbseq: A Python package for perturbational single-cell data analysis.

Support

For questions and support:

• Open an issue on GitHub
• Email: aidanw@arcinstitute.org

License

This project is licensed under the MIT License - see the LICENSE file for details.