perturbvi 0.2.1

Infer regulatory modules through informative latent component model in the single-cell Perturb-seq data

perturbVI

perturbvi is a scalable approach to infer regulatory modules through informative latent component model in the single-cell Perturb-seq data.

Installation | Example | Notes | Version | Support | Other Software

---

Installation

# install perturbvi
uv pip install perturbvi

# help
perturbvi --help

Get Started with perturbvi

Perform inference using SuSiE PCA to find the regulatory modules from CRISPR perturbation data

perturbvi <matrix> <guide> <z_dim> <l_dim> <tau> -o=<out_dir> --verbose

Arguments

• matrix: Path to the experiment CSV file.
• guide: Path to the guide CSV file.
• z_dim: Number of latent factors, Z dim (12).
• l_dim: Number of single effects, L dim (400).
• tau: Residual precision, Tau (800).
• out_dir: Specifies the output directory path.
• --verbose: For logging (Optional).

Example Usage

perturbvi luhmes_exp.csv luhmes_G.csv 12 400 800 -o=results --verbose

This will save all the files (including params.pkl) in results folder.

For luhmes results, we can analyze it:

import perturbvi

guide_path="luhmes_G.csv"
gene_symbol_path = "luhmes_gene_symbol.csv"

OUTPUT_DIR = "results"

info = perturbvi.utils.luhmes_analysis(
  OUTPUT_DIR, 
  guide_path, 
  gene_symbol_path
)

# number of degs per w from PIP
print(info["num_deg_per_w"])

# number of degs per perturbed gene from LFSR
print(info["num_deg_per_perturbed_gene"])

Notes

• perturbvi uses JAX with Just In Time compilation to achieve high-speed computation. However, there are some issues for JAX with Mac M1 chip. To solve this, users need to initiate conda using miniforge, and then install perturbvi using pip in the desired environment.

Version History

TBD

Support

Please report any bugs or feature requests in the Issue Tracker. If users have any questions or comments, please contact Dong Yuan ([e-mail]) and Nicholas Mancuso ([e-mail]).

Other Software

Feel free to use other software developed by Mancuso Lab:

• SuShiE: a Bayesian fine-mapping framework for molecular QTL data across multiple ancestries.
• MA-FOCUS: a Bayesian fine-mapping framework using TWAS statistics across multiple ancestries to identify the causal genes for complex traits.
• SuSiE-PCA: a scalable Bayesian variable selection technique for sparse principal component analysis
• twas_sim: a Python software to simulate TWAS statistics.
• FactorGo: a scalable variational factor analysis model that learns pleiotropic factors from GWAS summary statistics.
• HAMSTA: a Python software to estimate heritability explained by local ancestry data from admixture mapping summary statistics.

---

perturbvi is distributed under the terms of the MIT license.

---

This project has been set up using Hatch. For details and usage information on Hatch see https://github.com/pypa/hatch.