Dynamic dysregulation of the promoter DNA methylome is a signature of cancer. However, comprehensive understandings about how the DNA methylome is incorporated in the transcriptional regulation circuitry and involved in regulating the gene expression abnormality in cancers are still missing. We introduce an integrative analysis pipeline based on mutual information theory and tailored for the multi-omics profiling data in The Cancer Genome Atlas (TCGA) to systematically find dependencies of transcriptional regulation circuits on promoter CpG methylation profiles for each of 21 cancer types. By coupling transcription factors with CpG sites, this cancer type-specific transcriptional regulation circuitry recovers a significant layer of expression regulation for many cancer-related genes. The coupled CpG sites and transcription factors also serve as markers for classifications of cancer subtypes with different prognoses, suggesting physiological relevance of such regulation machinery recapitulated here. Our results therefore generate a resource for further studies of the epigenetic scheme in gene expression dysregulations in cancers.
Keywords: CpG dinucleotides; DNA methylation; MeTRN; cancer type-specificity; conditional mutual information; gene expression regulation; genome-wide analysis; pan-cancer analysis; transcriptional regulation; transcriptional regulation network.
Copyright © 2019 The Author(s). Published by Elsevier Inc. All rights reserved.