scDART -- Learning latent embedding of multi-modalsingle cell data and cross-modality relationshipsimultaneously

scDART v0.1.0

Zhang's Lab, Georgia Institute of Technology

Developed by Ziqi Zhang, Chengkai Yang

Description

scDART (single cell Deep learning model for ATAC-Seq and RNA-Seq Trajectory integration) is a scalable deep learning framework that embed the two data modalities of single cells, scRNA-seq and scATAC-seq data, into a shared low-dimensional latent space while preserving cell trajectory structures. Furthermore, scDART learns a nonlinear function represented by a neural network encoding the cross-modality relationship simultaneously when learning the latent space representations of the integrated dataset.

The preprint is available on Genome Biology: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-022-02706-x

Dependencies

Pytorch >= 1.5.0

numpy >= 1.18.2

scipy >= 1.4.1

pandas >= 1.0.3

sklearn >= 0.22.1

seaborn >= 0.10.0

Installation

Clone the repository with

git clone https://github.com/PeterZZQ/scDART.git

And run

pip install .

Uninstall using

pip uninstall scdart

Usage

See Example/demo.ipynb.

Contents

• scDART/ contains the python code for the package
• data/ contains the sample simulated dataset.
• Example/ contains the demo code of scDART.

Results in the preprint

The benchmark code, data and results are available through: https:github.com/PeterZZQ/scDART_test

The script for data simulation can be found through: https://github.com/PeterZZQ/Symsim2

Cite

Zhang, Ziqi, Chengkai Yang, and Xiuwei Zhang. "Learning latent embedding of multi-modal single cell data and cross-modality relationship simultaneously." bioRxiv (2021).