This repository contains a workflow for inference of transcriptional regulatory networks (TRNs) from gene expression data and prior information, as described in:
From gene expression data and tables of prior information, the example Th17 workflow can be used to infer a TRN using modified LASSO-StARS, and relies upon GlmNet in MATLAB to solve the LASSO. Workflow also includes TRN model evaluation based on precision-recall and ROC.
The resulting network can be visualized with TRN visualization software: jp_gene_viz.
Additional workflows are provided for:
- Construct a prior transcriptional regulatory network from ATAC-seq data
- TF-TF module analysis: Discovery of TFs that co-regulate gene pathways
- Identify "core" TF regulators for a subset of conditions or celltypes in the gene expression dataset
- Gene-set enrichment analysis (GSEA) of a TF's positive and repressed target genes
- Visualize TF degree, positive and negative edges, and overlap with TF-gene interactions in the prior
- Out-of-sample gene expression prediction, including calculation of R2pred
- Modeling of time-series gene expression with linear differential equations, as in Bonneau et al. (2006) Genome Biology
NOTE: For Mac users of MATLAB 2016b or later versions, you might need to install gfortran. We recommend the following: Install Homebrew, and then install gfortran with the Terminal command: "brew cask install gfortran".