Biostatistics Weekly Seminar

Genomewide analysis of DNA methylation patterns associated with breast cancer risk

Fei Ye, PhD
Vanderbilt University School of Medicine

Studies have suggested that genetic and epigenetic factors both play an important role in breast cancer etiology. DNA methylation is one of the most frequent and well-characterized epigenetic modifications, reflects at the molecular level a wide range of environmental exposures and genetic influences. Investigating the relationship between DNA methylation levels across the genome and breast cancer risk may discover novel biomarkers for the disease. Directly profiling the genome-wide methylation in breast tissues in a large population-based study, however, would be extremely costly and inefficient. Recently, studies have shown that DNA from blood samples could be used to reliably identify methylation markers associated with human diseases, including breast cancer, using a method known as methylation-wide association study. Current practice often uses elastic net regression, a hybrid regularization that blends both penalization of the L1 and L2-norms, to predict methylation levels from high-density genotyping data, and then test for associations between predicted methylation levels and disease outcome using “single-marker” type of analyses. One limitation of the association analysis part of this approach is the lack of acknowledging two important properties of methylation data: first, methylation levels are not independent, but often strongly related to each other; second, the number of candidate markers often greatly exceeds the number of subjects. In our study we conducted a genomewide analysis of DNA methylation patterns using elastic net regularization to predict methylation levels with genetic variants, then evaluate associations of predicted methylation markers with breast cancer risk via elastic net regression for binary classification in a large-scale case-control study. Elastic net regularization allows us to efficiently identify groups of potentially related genetic variants that explain most variation in the methylation levels. Similarly, in the association analysis, where interpretation of a complex set of correlated methylation markers is necessary, elastic net for binary classification uses the structure of the correlated methylation data to identify the combinations of methylation markers together contribute most to the separation of breast cancer cases and controls. Identified methylation markers will be validated in an independent set of samples with directly measured methylation levels and genotyping data.

MRBIII, Room 1220
12 December 2018

Speaker Itinerary

Topic revision: r1 - 10 Dec 2018, TawannaPeters

This site is powered by FoswikiCopyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback