Changelog

Version 0.1.0 (2025-01-26)

Initial release of scHopfield.

Features

Core Functionality

  • Sigmoid function fitting to gene expression distributions

  • Network inference from RNA velocity using gradient descent

  • Energy landscape computation and decomposition

  • GPU acceleration support for training and analysis

Network Analysis

  • Network centrality metrics (degree, betweenness, eigenvector)

  • Eigenvalue decomposition of interaction matrices

  • Network comparison across cell types

  • GRN visualization with customizable layouts

Stability Analysis

  • Jacobian matrix computation for all cells

  • Eigenvalue analysis for stability assessment

  • Rotational component analysis

  • Partial derivative computation for gene pairs

  • HDF5 storage for large Jacobian matrices

Visualization

  • Energy landscape plots

  • Interaction matrix heatmaps

  • GRN network graphs

  • Jacobian eigenvalue spectra

  • Centrality rankings and comparisons

  • Correlation scatter plots

Dynamics Simulation

  • ODE integration for gene expression trajectories

  • Perturbation experiments (knockouts, overexpression)

  • Trajectory visualization

Documentation

  • Complete API reference with numpy-style docstrings

  • User guide with detailed tutorials

  • ReadTheDocs integration

  • Example notebooks

API

  • scHopfield.pp - Preprocessing

  • scHopfield.inf - Network inference

  • scHopfield.tl - Analysis tools

  • scHopfield.pl - Plotting

  • scHopfield.dyn - Dynamics simulation

Dependencies

  • Core: numpy, scipy, pandas, matplotlib, anndata, scanpy, torch, networkx

  • Optional: seaborn, python-igraph, dynamo-release

Future Releases

Planned features for future versions:

  • More example notebooks with real datasets

  • Additional network analysis metrics

  • Enhanced perturbation analysis

  • Integration with trajectory inference tools

  • Performance optimizations

  • Additional visualization options