Background

Differential expression of genes across different tissues has been thought to play an important role in shaping human diseases (Oliva et al., 2020). An increasing body of studies has begun to characterize how genes are co-expressed in tissues, to rewire transcriptional regulatory networks as key molecular mechanisms underlying human disease (Saha et al., 2017; Malatras et al., 2020; Consortium, 2020). However, existing approaches for reconstructing gene regulatory networks are limited in capturing the full properties of gene–gene interactions, which are essential for a mechanistic understanding of disease etiology. For example, correlation-based approaches can only estimate the strength of gene–gene interactions, failing to identify the causality of the interactions, whereas Bayesian networks can identify the causality, but cannot characterize the sign of the interactions and feedback cycles. All these properties can be recovered using dynamic modeling of gene expression. However, it is impossible and ethically impermissible to collect temporal transcriptional data from human tissues. We have developed a series of quasi-dynamic models that can identify the aforementioned properties in gene–gene interactions from static data (Chen et al., 2019; Griffin et al., 2020; Wu and Jiang 2021; Chen et al., 2021). More recently, we leveraged these models to recover tissue-specific and pseudo-dynamic gene regulatory networks across the spectrum of disease risk (Chen et al., 2022). In this article, we develop a software pipeline called D isease R isk-associated pseudo-D ynamic Networks Bio-Pro tocol (DRDNetPro). This pipeline provides a detailed tutorial, enabling researchers to reconstruct pseudo-dynamic networks from static gene expression data, which helps to identify context-specific and personalized networks, and further the mechanistic understanding of diseases. The software and detailed tutorials with illustrative data examples can be found in our GitHub repository (https://chencxxy28.github.io/DRDNetPro/articles/NAME-OF-VIGNETTE.html).

Equipment

Computational requirements

1. The users need to prepare a desktop/laptop computer with R version 4.1.0 or above installed.

   
   

Install packages (The summary of the tutorial in the GitHub page repository)

1. Run the function “install.packages()” to install the required packages, including “pROC”, “np”, “splines2”, “grpreg”, “Matrix”, “igraph”, “ggplot2”, and the optional packages “ranger”, “XGBoost”, and “dplyr”.

2. Run the function “devtools::install_github()” to install the package “DRDNetPro”. Run the function “install.packages()” to install the package “devtools”, if it has not been installed yet.

Software

Software list

1. DRDNetPro (Chixiang Chen, https://github.com/chencxxy28/DRDNetPro)

2. graph (The igraph core team, https://igraph.org/r/)

3. ggplot2 (Hadley Wickhan, https://ggplot2.tidyverse.org/)

   
   

Datasets for demonstration

The demo data for method illustration can be downloaded from the website https://chencxxy28.github.io/DRDNetPro/articles/web/data.html, which is from the GTEx project originally. Gene expression data were collected from blood vessels and de-identified subsamples of phenotype information from donors.

Procedure