PORCELAN: Integrating representation learning, permutation, and optimization to detect lineage-related gene expression patterns

This repository contains code for the paper "Integrating representation learning, permutation, and optimization to detect lineage-related gene expression patterns". We developed Permutation, Optimization, and Representation learning based single Cell gene Expression and Lineage ANalysis (PORCELAN) to identify lineage-informative genes or subtrees where lineage and expression are tightly coupled:

Repository overview

• data contains jupyter notebooks for downloading, simulating, and pre-processing the datasets used in the paper's results. For convenience, we also provide pre-processed data files in data/preprocessed. See data/README.md for further details.
• figure_notebooks contains jupyter notebooks to reproduce the paper's main and supplemental figures. Most results can be reproduced in a few seconds or minutes but we also provide the data files for the results displayed in the figures in results for convenience. See figure_notebooks/README.md for further details.
• tutorial contains a jupyter notebook that walks the user through applying all componenets of PORCELAN to an example tumor. See tutorial/README.md for further details.

Dependencies:

Python:

This repository was developed using Python 3.8. You can use Conda to create a virtual environment for a specific Python version. Additional required packages are listed in requirements.txt and can be installed using the following command:

pip install -r requirements.txt

Installing dependencies can take a few minutes or up to an hour dependending on how many packages need to be downloaded rather than reusing cached versions.

R:

We only use R to simulate lineage-resolved gene expression data with TedSim (installation instructions).

Operating system and hardware:

We tested this code on a machine running Rocky Linux 8.8 (Green Obsidian) and equipped with an NVIDIA RTX A6000 GPU.

Citation

TODO