RedeStiff

RedeStiff is a R-based computational software for eastimation of tissue stiffness score based on ligand-receptors interaction from RNA-seq data. This tool was tested on various RNA-seq datasets across several sequencing platforms including bulk RNA-seq, Visium, slide-RNA-seq, Xenium and so on. RedeStiff was tested on Windows 11 and CentOS 7. Through RedeStiff, users can not only predict tissue stiffness score from RNA-seq data but also identify highly contributed ligand-receptors to tissue stiffness. RedeStiff could probably provide new molecular targeted therapy for different diseases.

Installtion

install.packages("devtools")
library(devtools)
devtools::install_github("gancao/RedeStiff")

Usage

Before running RedeStiff, the users require downloading the trained ligand-receptor (LR) weight list file from https://github.com/gancao/RedeStiff and providing expression matrix file in TPM values (Row:Gene, Column: Sample). In addition, output folder (result_dir) and the prefix name of reuslt files are required to input to RedeStiff.
step1: download LR weight list file

wget -c https://github.com/gancao/RedeStiff/blob/main/files/LR_weights_final.txt

step2: run RedeStiff

library(RedeStiff)
analyse_stiffness(train_LR_weight_file,exp_file,result_dir,prefix='RedeStiff',parallel_flag='yes',cores=50)

Alternatively, you can run RedeStiff by three stages:

exp_count <- read.delim(exp_file,header=T,row.names=1,check.names=F)

#step1:calculate LR score matrix
LR_sample_matrix <- calculate_LR_score(exp_count,train_LR_weight_file,parallel_flag='yes',cores=50)
sample_LR_matrix <- t(LR_sample_matrix)
save_sample_lr_matrix_file <- paste(result_dir,'/',prefix,'_sample_LR_score.txt',sep='')
write.table(sample_LR_matrix,save_sample_lr_matrix_file,quote=F,sep="\t")

LR_weights <- read.delim(train_LR_weight_file,header=T,check.names = F)
lr_weights <- LR_weights[,2]
names(lr_weights) <- as.character(LR_weights[,1])

#step2:calculate stiffness
data_result <- calculate_stiffness(save_sample_lr_matrix_file,lr_weights,result_dir,prefix)
data_result_file <- paste(result_dir,'/',prefix,'_stiffness_score.txt',sep='')

#step3:calculate LR contribution
data_result <- read.delim(data_result_file,header=T)
predict_LR_contribution(save_sample_lr_matrix_file,lr_weights,data_result,result_dir,prefix)

The required parameters are listed as follows:

train_LR_weight_file: The trained LR weight table file in two colums, the first column represents LR pairs and the second column represents corresponding weight of the LR.
exp_file: Gene expression matrix file of samples that user input (Row:Gene, Column: Sample).
result_dir: The ouput folder of all results.
prefix: the prefix name of reuslt files.

The optinal parameters are listed as follows:

parallel_flag: "‘yes’" (default) or "‘no’", "‘yes’" denotes multiple cores and "‘no’" denotes single core for calculaing LR score matrix;
cores: Number of cores to be used when doing parallel computing, default is "50".

Output

The contents of the output directory include three files:

<prefix>_sample_LR_score.txt: A matrix file that saves ligand-receptor scores where row represent samples, columns represent ligand-receptors and the values represent the product of the ligand and receptor TPM values in the sample.
<prefix>_stiffness.txt: A table file in tab-delimited format that saves the predicted stiffness of each sample. The first column represents sample identifiers given by tpm_file and the second column represents the corresponding stiffness value.
<prefix>_sample_LR_contribution.txt: A matrix file that saves ligand-receptor contribution scores where row represent samples, columns represent ligand-receptors and the values represent the corresponding contribution score.

Others

We provide test files (GSE145429_tpm.txt) under "files". In addition, other files, codes and figures can be downloaded from https://zenodo.org/records/11514581.