gwaybio / brca_lowstage_dmgrs

The repository stores the analytical code for the pre-release of the manuscript titled "Tumor subtype and cell type independent DNA methylation alternations reveals twelve genes associated with low stage breast carcinoma".

Cell-type (and subtype) independent DNA methylation alterations in breast cancer

Titus, A.J., Way, G., Johnson, K., Christensen, B. 2016, Manuscript in preparation

Summary

The following repository contains all scripts required to reproduce an analysis of low stage invasive breast carcinoma investigating similarities in DNA methylation measured by The Cancer Genome Atlas on the Illumina 450k platform. At the core of the analysis is a reference free adjustment of cell type proportion (Houseman, Molitor, and Marsit. Bioinformatics 2014) on each PAM50 subtype stratified by low and high stages. After the adjustment, we observe key gene regions of differential methylation in common to all PAM50 subtypes in the low stage. We also validate these findings in a Validation set (Yang et al. Genome Biology 2015).

Contact

For all code related questions please file a GitHub issue

Questions regarding the analysis or other correspondance should be directed to: [email protected]

Analysis

All scripts are intended to be run in order, as defined in run_pipeline.sh. The bash script can be run directly to reproduce the pipeline but this is not recommended since some steps are computationally intensive. Instead, to successfully reproduce this analysis, we recommend following the bash script line by line.

The folder structure is labelled to facilitate easy step-wise navigation, as are the Scripts/ folders in each parent folder. All scripts should be run from the top directory in the repository. For example:

# Place downloaded IDAT files in this folder
IDAT_loc="../../Documents/mdata/TCGAbreast_idat/"

# Run preprocessing script 
R --no-save --args "I.Data_Processing/Data/TCGA_BRCA" $IDAT_loc "nosummary" < \
I.Data_Processing/Scripts/A.preprocess_minfi.R

Data

Data is not stored directly in this repository and should be downloaded according to the run_pipeline.sh script.

Acknowledgements

This work was supported by the Institute for Quantitative Biomedical Sciences and two grants P20GM104416 and R01DE02277 (BCC)