This repository presents an illustrative application of the DRAGON algorithm (https://arxiv.org/abs/2104.01690) to promoter methylation and gene expression in breast cancer data from TCGA. All of the data are downloaded directly from TCGA; running these scripts will require that you have approximately 20G of storage available.
Clone this repository. To set up the Python and R tools you will need, install and activate the conda environment in dragon_env.yml and install R packages using installPackages.R.
From the base directory in the repository, run the three shell scripts sequentially:
./runPreprocessing.sh
./runProcessing.sh
./runPostprocessing.sh
Reference:
In src/preprocessing, you will find the scripts used to pull TCGA data, to map methylation probes to TF promoter regions, and to merge phenotypic, methylation, and gene expression data.
In src/processing, you will find the scripts used to clean the promoter-level methylation values and the gene expression data along with the scripts to run DRAGON.
In src/postprocessing, you will find the scripts used to analyze the DRAGON results and make the tables and figures for these sections of the paper.
This analysis was run on Linux. On other OS, there is sometimes a difference in GenomicDataCommons functionality: the attribute file_name of the GDC manifest may be filename (no underscore) and you will need to change this in the scripts.