stMLnet

What is stMLnet?

`stMLnet` is a tool to infer spatial intercellular communication and multilayer signaling regulations from `spatial transcriptomic data (ST)` by quantifying distance-weighted ligand–receptor (LR) signaling activity based on diffusion and mass action models and mapping it to intracellular targets. stMLnet can infer, quifity, and visualize both intercelluar communitations and the intracelluar gene regulatory network from ST data. stMLnet allows:
* to construct a multilayer signaling network, infer LR signling activate, and predicte LR-target gene regulation
* to leverage spatial information in the ST data to quantify intercellular signaling activity and connect extracellular signals to intracellular gene expression
* to visualize inter- and intra-cellular signaling networks and functions associated with cellular communications and molecular regulations

We also provide the R code used for collection and integration of prior databases, detailed in stMLnet-AnalysisCode repository

Package Structure

The repository is centered around the R module:

• creat_multilayer_network contains the scripts to create mulitlayer signling network
  
• calculate_signal_activity contains the scripts to obtain the upstream paired signaling activity and downstream target gene expression based on the mulitlayer signling network
  
• calculate_signal_importance contains the scripts to calculate the upstream signal pairs or signals importance in the multilay signal network of cell communication
  
• visualize_cell_communication contains the scripts to visualize cell-cell interations
  
• check_feedback_loop contains the scripts to select multicellular feedback circuits
  

Usage

(1) Install related dependencies:

(1.1) Manual installation

 pkgs <- c("R.utils",
      'reshape2','stringr','dplyr', # for data preprocessing
      'caret','doParallel','snow','foreach',"doSNOW","ranger", # for quantitative model
      'ggplot2','ggsci', 'plotrix','ggalluvial','ggraph','igraph' # for visualization
           )
 for (pkg in pkgs) {
    if (!requireNamespace(pkg)) {
        install.packages(pkg, repos = 'https://cloud.r-project.org')
    }
 }

pkgs <- c('Seurat','org.Hs.eg.db')
for (pkg in pkgs) {
  if (!requireNamespace(pkg)) {
  BiocManager::install(pkg)
 }
}

Or

(1.2) Dock image environment

Alternatively, if you have problems installing the environment manually, you can also choose to install the dependent environment via dockfile:

   # Bash
   # built a docker image
   # ensure that dockerfile and postInstall are in the same path
   docker build -f Dockerfile -t stMLnetEnv:0.1 .
   # Run docker image
   docker run -it stMLnetEnv:0.1 /bin/bash

(2) Install stMLnet package

After building dependent environment, you can

(2.1) download stMLnet from github:

   git clone https://github.com/SunXQlab/stMLnet.git

and then install stMLnet from local:

   install.packages("path/to/stMLnet/stMLnet_0.1.2.tar.gz", repos = NULL, type = "source")
   library(stMLnet)

Or

(2.2) you can directly install stMLnet from github:

   devtools::install_github("SunXQlab/stMLnet")
   library(stMLnet)

To learn how to use this tool, check Tutorial of stMLnet.Rmd. This tutorial shows the installation and application of stMLnet in the demo dataset, which can be download from here. It will take about 15 mins to run this demo (excluding environment installation) mainly depending on the parameter setting in the quantitative analysis step.

Examples and Reproducibility

All the examples and the reproducibility codes for the plots in the paper could be found in the stMLnet-AnalysisCode repository which includes:

• prior_knowledge contains the code used for collection and integration of prior databases
  
• apply_in_simu contains the code to reproduce the simulation study of stMLnet
  
• apply_in_scST contains the code to reproduce the plot and detailed analysis of the three single-cell resolution ST datasets
  
• apply_in_COVID19 contains the code to reproduce plots and detailed analysis of the COVID-19 ST dataset
  
• benchmark contains the code to reproduce plots and benchmarking
  
• code contains all functions of stMLnet to analysis cell-cell interactions
  

See detials therein.

We also provide a web-based application to demonstrate the functionality and visualization of stMLnet, available at http://net.stmlnet.top.

Contact

If you have questions or suggestions for imrpoving stMLnet, please contact Xiaoqiang Sun via sunxq6@mail.sysu.edu.cn.