MethylModes: computationally efficient detection of multimodal distributions in DNA methylation data

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• DNA methylation patterns can exhibit multimodal distributions due to genetic variation, environmental factors, or technical artifacts.
• Identifying these multimodal distributions is crucial for accurate DNA methylation analysis and quality control.
• Such distributions can introduce confounding variables in epigenetic studies.

Purpose of the Study:

• To introduce MethylModes, an R package and Shiny application designed for detecting multimodal distributions in human DNA methylation data.
• To provide a user-friendly tool for integrating into existing quality control pipelines for array-based DNA methylation data.
• To offer an efficient and scalable solution for genome-wide analysis of DNA methylation patterns.

Main Methods:

• Utilizes kernel smoothing of probe-level DNA methylation data to identify the number and location of peaks.
• The algorithm is designed for parallel processing across probes, enabling efficient genome-scale analysis.
• Implemented as an R package and Shiny application, accessible via GitHub.

Main Results:

• MethylModes successfully identifies multimodal distributions in DNA methylation data at individual CpG sites.
• The tool can be readily incorporated into quality control workflows for large-scale epigenomic studies.
• Demonstrated utility through case studies in the Health and Retirement Study and the Airwave Health Monitoring Study.

Conclusions:

• MethylModes provides a robust method for identifying multimodal DNA methylation patterns, crucial for robust epigenetic research.
• The package enhances the quality control of array-based DNA methylation data analysis.
• Its efficient and scalable design supports genome-wide applications and integration into existing bioinformatics pipelines.