Method to detect differentially methylated loci with case-control designs using Illumina arrays

Area of Science:

• Genomics
• Cancer Biology
• Epigenetics

Background:

• Cancer arises from accumulated genetic and epigenetic alterations.
• DNA methylation is a key epigenetic modification vital for development, often dysregulated in cancer.
• High-throughput technologies generate vast amounts of genomic methylation data, necessitating advanced analytical methods.

Purpose of the Study:

• To develop a sophisticated statistical method for analyzing complex DNA methylation data.
• To identify differentially methylated loci between cancer cases and controls using Illumina array data.
• To address the limitations of existing statistical approaches for large-scale methylation studies.

Main Methods:

• Developed a likelihood-based Uniform-Normal-mixture model.
• Modeled methylation data using three components: unmethylated, completely methylated, and partially methylated loci.
• Employed a three-component mixture model with Uniform and truncated normal distributions.

Main Results:

• Demonstrated the feasibility and statistical power of the proposed Uniform-Normal-mixture model through simulations.
• Successfully identified differentially methylated loci between case and control groups.
• Applied the method to ovarian cancer data, uncovering methylation loci missed by existing techniques.

Conclusions:

• The Uniform-Normal-mixture model provides a powerful and effective approach for analyzing genome-wide DNA methylation data.
• This method enhances the identification of key methylation changes associated with cancer.
• The developed model offers a valuable tool for advancing cancer epigenetics research.