ENTRAIN is an R and python package for analysis of single-cell RNA sequencing data.
The purpose of ENTRAIN is to predict how the cellular environment shapes cellular differentiation, and quantify the relative contribution of cell-cell communication versus cell-intrinsic mechanisms towards cell fate specification. If you have a dataset of cells that lie along a differentiation trajectory, and/or exhibit robust RNA velocity dynamics, then ENTRAIN will help you generate hypotheses for identifying the environmental signals that are driving your observed dynamics.
As input, ENTRAIN takes single-cell RNA sequencing data (and optionally, RNA velocities), and integrates with the regulatory databases to predict how cell-cell communication can play a role in cellular maturation and decision making.
ENTRAIN is based on, and can be considered a unification, of three previously separate streams of single-cell analysis: Trajectory Analysis (with monocle3), cell-cell communication/regulation (with NicheNet), and RNA Velocity (with scvelo). Furthermore, ENTRAIN works directly with popular single-cell data types, namely Seurat objects and anndata objects.
Installation
ENTRAIN is available through Github, using devtools:
if (!require(devtools)){
install.packages("devtools")
}
devtools::install_github(c("satijalab/seurat-wrappers", "theimagelab/entrain"))-
seurat-wrappersinstallation sometimes fails because it cannot findR.utils. If that occurs, try runninginstall.packages("R.utils")first.
ENTRAIN is also available as a python package through pypi:
conda install squidpy scanpy rpy2 scvelo adjusttext r-randomforest
pip install tangram-sc
pip install pypath-omnipath
pip install entrainConfiguring Reticulate (Entrain-Velocity in R only).
Python users can skip this. For users wishing to run ENTRAIN analysis on RNA velocity data, you can do this in both R and Python. If you wish to do this in R, we we require you to install reticulate, an R interface to python for natively manipulating RNA velocity data created by scvelo. The following code installs reticulate, python (via Miniconda), and a few python packages into a standalone conda virtual environment called r-reticulate-sc. This conda environment will be isolated from any other python environments on your system, preventing package conflicts.
reticulate::install_miniconda()
reticulate::conda_create("r-reticulate-sc")
reticulate::conda_install("r-reticulate-sc", "scanpy", "leidenalg", "python-igraph", "adjustText", channel="conda-forge")
reticulate::conda_install("r-reticulate-sc", "scvelo", channel="bioconda")
reticulate::use_condaenv("r-reticulate-sc")Using Entrain
ENTRAIN has three types of analysis. Read one of the example vignettes to learn how to use Entrain.