HopfieldModel 1.0.1

A Python package to model single-cell data using Hopfield networks

HopfieldModel

Author: Ashish Phal

BIOEN537: Computational Systems Biology. University of Washington, Seattle.

A Python package designed to model single-cell data using Hopfield networks, enabling energy-based analysis of cellular transitions.

License: MIT
Current version: 1.0.0
Last updated: 2024-12-10

Background

Understanding cellular differentiation and transitions is crucial in developmental biology and regenerative medicine. Hopfield networks provide a framework for modeling these processes by representing gene expression states as memory patterns and computing energy landscapes that describe transitions between cell states.

The HopfieldModel package facilitates these analyses with tools to normalize gene expression data, select highly variable genes, compute cell-specific Hopfield energy, and visualize cellular transitions. The package is intended for researchers and students working with single-cell RNA sequencing data, offering a streamlined and accessible approach to study cell states and transitions.

---

Installation

Package Dependencies

This package requires Python 3.7 or higher and the following Python packages:

• numpy: Numerical computing
• matplotlib: Data visualization
• scikit-learn: Machine learning and PCA analysis
• scanpy: Single-cell analysis
• seaborn: Statistical data visualization
• scipy: Scientific computing

These dependencies will be automatically installed when you install the package using pip.

Installing the Package

To install the package, run the following command:

pip install HopfieldModel

Example Visualizations

1. Transition Energy Plot

This plot shows the transition of cells between two specified types along PC1, with their Hopfield energy.

---

2. PCA Visualization

Cells are projected into PCA space, with each point colored by its Hopfield energy. Different colors represent distinct cell types.

---

3. Gene Transition Matrix

This heatmap displays the gene state changes between two cell types, clustered hierarchically. The clustering highlights key genes involved in the transition.

---

4. Hopfield Energy Boxplot

This boxplot shows the distribution of Hopfield energy across different cell types, enabling a comparison of differentiation potency.