Covariance Inference for Perturbation and High-dimensional Expression Response

CIPHER is a tool designed for analyzing covariance structures in high-dimensional gene expression data, particularly in response to perturbations.

• Covariance inference for gene expression datasets
• Designed for high-dimensional data
• Suitable for perturbation experiments

# Example installation command
bash .init_conda.sh
mamba activate .conda/cipher

from src.r2 import full_analysis_with_nulls_soft_and_plots
# Example usage of CIPHER module
adata_path = 'path/to/your/perturb-seq-adata.h5ad'
output_dir = 'path/to/output_dir'
full_analysis_with_nulls_soft_and_plots(adata_path, save_dir=output_dir)

Notebooks

Notebooks to reproduce each figure of the paper can be found in the notebooks directory. All notebooks work with the supplied conda environment.

notebooks/LR_fig2.ipynb

• Fig. 2 (All) notebooks/LR_fig3_R2_hist.ipynb
• Fig. 3 (A-M) notebooks/LF_double_pert_R2_and_inference.ipynb
• Fig. 3 (N, O, P)
• Fig. 4 H notebooks/LR_fig3_cross_dataset.ipynb
• Fig. 3 Q, R notebooks/LR_fig4.ipynb
• Fig. 4 (A-G) notebooks/LR_fig5_TRADE_and_EGENES.ipynb
• Fig. 5

Link to google drive with source data https://drive.google.com/drive/folders/1HYEDb_7tbaO0XvQ_36t2WuA2vgYp3YX5

Method benchmarks Benchmarking notebooks can be found in the benchmarks/ folder

MIT License