CauFinder Documentation

Introduction

CauFinder is a powerful framework designed to identify and control causal regulators of cell-state and phenotype transitions. By leveraging causal disentanglement modeling and network control, CauFinder provides insights into key regulatory mechanisms involved in various biological processes such as cell differentiation, cancer transdifferentiation, and drug resistance transitions.

CauFinder excels in distinguishing causal factors from spurious ones, ensuring precise control of state transitions. This documentation will guide you through the setup, usage, and application of CauFinder in different biological contexts.

Overview

The image above provides a high-level overview of the CauFinder framework, illustrating its core components and workflow.

Contents

• Installation
  • Clone the repository
  • Set up conda environment
  • Install additional dependencies
  • Optional: Install GUROBI
• Simulated Data Benchmark
  • Importing necessary libraries
  • Benchmark main function
  • Run benchmark
• Human LUAS
  • Import necessary packages for CauFinder
  • Set path for read files and save output
  • Run CauFinder
  • Plot PCA with probabilities
  • Run state transition
  • Drawing state transition score on PCA
• Mouse LUAS
  • Import necessary packages for CauFinder
  • Set path for read files and save output
  • Run CauFinder (one seed example)
  • Loading exist model and driver information (output of CausalFinder)
  • Run state transition
  • Drawing state transition score on PCA
  • Draw 3D state transition plot for samples

References

If you use CauFinder in your research, please cite the following paper:

Chengming Zhang, Zexi Chen, Yuanxiang Miao, Yun Xue, Deyu Cai, Weifeng Guo, Hongbin Ji, Kazuyuki Aihara, Luonan Chen. “Steering cell-state and phenotype transitions by causal disentanglement learning.” bioRxiv (2024): [https://doi.org/10.1101/2024.08.16.607277](https://doi.org/10.1101/2024.08.16.607277).