Testing differentially methylated regions through functional principal component analysis

Area of Science:

• Epigenetics and Genomics
• Computational Biology
• Statistical Bioinformatics

Background:

• DNA methylation is a key epigenetic regulator in biological processes and disease.
• Current methods often analyze individual CpG sites, necessitating post-hoc regional analysis.
• Regional methylation analysis faces challenges due to site-level variability in Next-Generation Sequencing (NGS) data and limitations of parametric models.

Purpose of the Study:

• To develop a robust nonparametric statistical approach for detecting differentially methylated regions (DMRs).
• To address the limitations of existing methods in handling noisy, site-level methylation data for regional analysis.
• To evaluate the performance of the novel approach against established methods.

Main Methods:

• Development of a nonparametric method utilizing functional principal component analysis (FPCA).
• Application of the FPCA-based approach to detect predefined differentially methylated regions (DMRs).
• Comparative performance analysis using both simulated and real-world Next-Generation Sequencing (NGS) methylation data.

Main Results:

• The proposed FPCA-based nonparametric method demonstrates effectiveness in identifying DMRs.
• The approach successfully navigates the challenges posed by site-level methylation variability in NGS data.
• Performance evaluation indicates comparable or superior results to existing methods like GIFT and M3D.

Conclusions:

• Functional principal component analysis offers a powerful nonparametric framework for regional DNA methylation analysis.
• The developed method provides a valuable tool for robustly detecting differentially methylated regions (DMRs).
• This approach enhances the ability to identify biologically significant regional epigenetic alterations from complex sequencing data.