The association between copy number aberration, DNA methylation and gene expression in tumor samples

Abstract

We systematically studied the association between somatic copy number aberration (SCNA), DNA methylation and gene expression using -omic data from The Cancer Genome Atlas (TCGA) on six cancer types: breast cancer, colon cancer, glioblastoma, leukemia, lower-grade glioma and prostate cancer. A major challenge for such integrated study is that the association between DNA methylation and gene expression is severely confounded by tumor purity and cell type composition, which are often unobserved and difficult to estimate. To overcome this challenge, we developed a method to remove confounding effects by calculating the principal components that span the space of the latent factors. Another intriguing findings of our study is that there could be both positive and negative associations between SCNA and DNA methylation, while the CpGs with negative/positive associations with SCNA are often located around CpG islands/ocean, respectively. A joint study of SCNA, DNA methylation, and gene expression suggest that SCNA often affect DNA methylation and gene expression independently.

A pan-cancer atlas of transcriptional dependence on DNA methylation and copy number aberrations

“…These trends indicate that the mutual exclusivity in E-M and E-C associations may partially be explained by copy number aberrations being able to override modulating effects of methylation on gene expression. This would be in agreement with findings from Sun et al 2 in which they arrive at similar conclusions based on analyses of conditional independence in expression, methylation and copy number. The trend of mutual exclusivity should also be viewed in light of the earlier discussion, in which we questioned the role of E-M associations in driving genome wide transcriptomic variation within tumor types ( Supplementary Fig.…”

“…Changes in DNA methylation and copy number are frequently assumed to have a phenotypic effect. However, there is major variation, between genes and tumor types, in how strongly these features affect transcription, and consequently, their functional significance 1,2,3 . The relationship between genetic factors and gene expression is essential to understanding transcriptomic regulation and heterogeneity in cancer, yet it is not well characterized at the whole-genome or pancancer level.…”

“…4g). Taken together, these correlations raise questions about whether methylation drives genome-wide transcriptional dysregulation in cancer (as appears to be the case for copy number aberrations), fine-tunes transcriptional patterns at a smaller scale, or mostly plays a passive role in maintaining transcriptional states 2,11 . Interestingly, GO-term analyses 12,13,14 revealed considerable enrichment of immune-related processes in the 20% most highly methylationassociated genes in most tumor types ( Supplementary Data 4).…”

IMIX: A multivariate mixture model approach to integrative analysis of multiple types of omics data

“…Based on the principles of molecular biology and previous publications (Sun et al, 2018;Wei and Pan, 2012), when a gene is associated with an outcome through, for example, DNA methylation and gene expression, the two events can be correlated. In particular, Wei and Pan (2012) developed an integrative genomic method to improve the power of a twocomponent mixture model by considering the possible correlations between three data types.…”

“…A common strategy is to assess the associations between genes and an outcome separately for each data type using Bonferroni correction or the Benjamini-Hochberg false discovery rate (BH-FDR) procedure (Benjamini and Hochberg, 1995) to adjust for multiple hypothesis testing. However, study on The Cancer Genome Atlas (TCGA) has found that omics data, such as gene expression, DNA methylation, and CNV, have several different triangular dependence structures (Sun et al, 2018). Furthermore, it remains unclear whether the correlation structures between data types vary according to the associations between different genes and the outcome of interest through those data types.…”