cRegulon

cRegulon is an optimization model to identify combinatorial regulon from single cell expression and chromatin accessibility data.

Introduction

This is cRegulon software: an optimization model to identify combinatorial regulon from single cell expression and chromatin accessibility data.

Requirements:

1. Python >=3.0 with packages: numpy, sklearn, and scipy
   
2. matlab >= 2021
3. Homer

Installing cRegulon with the following command:

wget https://github.com/fengzhanying/cRegulon/archive/master.zip
unzip master.zip
cd cRegulon-master
wget https://www.dropbox.com/s/0h1wxlu7iqheajo/cRegulon.tar.gz
tar -xzvf cRegulon.tar.gz

Step 1: single cell data preprocessing

The typic input file (RAd4_scRNA.txt) of scRNA-seq data is a gene by cell count matrix:

scRNA	Cell1	Cell2	Cell3
Gene1	5	0	3
Gene2	0	2	0
Gene3	1	0	0

The typic input file (RAd4_scATAC.txt) of scATAC-seq data is a peak by cell count matrix:

scATAC	Cell1	Cell2	Cell3	Cell4
Peak1	1	0	1	0
Peak2	0	1	0	1
Peak3	1	0	0	0

The peaks are in the format of "chr_start_end".
We run the following script to make the gene expression matrix and gene activity matrix:

source Preprocessing.sh RAd4

This process will produce gene expression file (RAd4_GE.txt) and gene activity file (RAd4_GA.txt)

Step 2: Constructing TF-TG regulatory network by pseudo-bulk strategy

With the input files are (RAd4_scRNA.txt) and (RAd4_scATAC.txt), we run the following script:

source PS_PECA.sh RAd4 mm10

This process will produce the TF-REs-TG triplets files (RAd4_network.txt) and TF-TG regulatory strength file (RAd4_TRS.txt).

Step 3: Constructing TF-TF combinatorial network

With the input TF-TG regulatory strength file (RAd4_TRS.txt), we run the following script:

source runCSI.sh RAd4

This will generate normalized TF-TG regulatory strength file (RAd4_TRS.txt) and TF-TF combinatorial network (RAd4_CSI.txt).

Step 4: Running cRegulon model

With the input of TF-TF combinatorial network (C.txt), normalized TF-TG regulatory strength matrix (R.txt), gene expression matrix (GE.txt), and gene activity matrix (GA.txt), we run the following cRegulon model:

source cRegulon.sh RAd4

This will output:

1. TF combinatorial effects in all cRegulons: X.txt
   
2. cRegulon combination coefficients for scRNA-seq: H1.txt
   
3. cRegulon combination coefficients for scATAC-seq: H2.txt
   
4. TF modules of cRegulons: TFs (*TF.txt) and TF pairs (*TFPair.txt).
5. Regulatory sub-network of each cRegulon: *SubNet.txt

Citation:

If you use cRegulon software or cRegulon associated concepts, please cite

Zhanying Feng, et al. Modeling combinatorial regulon from single cell gene expression and chromatin accessibility data. 2023.