This repository contains the scripts and notebooks for the preprint "Biophysical modeling with variational autoencoders for bimodal, single-cell RNA sequencing data". Although variational autoencoders can be treated as purely phenomenological and descriptive, without any explicit claims about the processes that generated data, it is possible to exploit implicit physics encoded in the mathematical formulation to model biophysical processes that gave rise to observations. By interpreting the scVI generative model as a description of a particular biophysical model, we can represent bivariate RNA distributions. We benchmark the implementation, biVI, on simulated and biological data.
BIVI/ contains all of the scripts used to implement biVI, while Manuscript/analysis/ contains all of the notebooks and scripts used to generate the manuscript figures and results. Example/ contains kb_pipeline.sh, a script that demonstrates how to align raw reads to a reference genome to obtain the unspliced/spliced count matrices necessary for biVI, and 'Demo.ipynb,` a Google Colaboratory notebook that processes the output matrices, train a biVI model, and visualize the results.
The biVI software can be installed as a standalone package using the following command:
pip3 install git+https://github.com/pachterlab/CGCCP_2023.git#subdirectory=BIVI .
If package dependencies cause installation issues, create a clean Conda environment and rerun the installation:
conda create --name biVI_env python==3.9
pip3 install git+https://github.com/pachterlab/CGCCP_2023.git#subdirectory=BIVI .
Installation takes one to several minutes on a standard laptop. Alternatively, biVI can be run in a Google Colab notebook, an example of which is given in Example/Demo.ipynb.