moflow 0.1.0

MoFlow is a deep learning framework for multi-omic RNA velocity modeling that extends the relay velocity model by incorporating chromatin accessibility.

MoFlow

MoFlow is a deep learning framework for multi-omic RNA velocity modeling that extends the relay velocity model (cellDancer) by incorporating chromatin accessibility.
By leveraging gene- and cell-specific kinetic parameters, MoFlow can jointly model chromatin accessibility, transcription, splicing, and degradation, enabling the study of transcriptional dynamics across diverse cell states.

---

Installation

Clone the repository and set up a conda environment:

git clone https://github.com/AriHong/MoFlow.git
cd MoFlow

conda create -n moflow python=3.7.0
conda activate moflow
pip install -r requirements.txt

---

Quick Start

We provide a demonstration notebook under notebooks/Demo.ipynb showing how to run MoFlow on a toy dataset and compute downstream scores. Also, the repository includes notebooks under notebooks/ for reproducing figures from the manuscript.

---

Acknowledgements

Portions of this code are adapted from the cellDancer repository:
https://github.com/GuangyuWangLab2021/cellDancer/

We thank the authors of cellDancer and MultiVelo for making their work publicly available.