Statistical methods for detecting differentially methylated regions based on MethylCap-seq data.

Area of Science:

• Epigenetics
• Genomics
• Bioinformatics

Background:

• DNA methylation is a key epigenetic regulator, with altered patterns linked to disease.
• Current methods for detecting differential methylation from MethylCap-seq data face challenges like high false-positive rates due to data sparsity and high dimensionality.
• Existing statistical approaches struggle with the correlation of signals between CpG sites.

Purpose of the Study:

• To review and evaluate the applicability of high-dimensional mean vector tests for identifying differentially methylated regions (DMRs).
• To compare the performance of mean vector tests against other established DMR detection methods.
• To provide recommendations for optimal statistical tests in DNA methylation analysis.

Main Methods:

• Utilized probabilistic approaches to distribute methylation signals at nucleotide resolution from MethylCap-seq data.
• Conducted comprehensive simulation studies to assess the performance of high-dimensional mean vector tests under various conditions.
• Compared mean vector tests with existing statistical methods for DMR detection.

Main Results:

• High-dimensional mean vector tests demonstrate superior performance in detecting differentially methylated regions (DMRs).
• These tests effectively address the challenges of 'curse of dimensionality' and signal sparsity inherent in methylation data.
• Identified significant enrichment of cancer-related canonical gene pathways in acute myeloid leukemia and ovarian cancer using mean vector tests.

Conclusions:

• High-dimensional mean vector tests offer a robust and accurate approach for detecting differentially methylated regions (DMRs).
• These methods are particularly effective in identifying biologically relevant pathways in cancer genomics.
• The study recommends mean vector tests as an optimal choice for analyzing DNA methylation data, especially in disease-associated studies.