Efficient gene set analysis for DNA methylation addressing probe dependency and bias.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene Set Enrichment Analysis (GSEA) is crucial for interpreting DNA methylation data by linking methylation sites to biological pathways.
• Existing GSEA methods face challenges with DNA methylation data, including probe dependency, probe number bias, and complex gene-probe mapping, leading to biased results and reduced power.
• These limitations hinder accurate biological interpretation and efficient computational analysis of methylation datasets.

Purpose of the Study:

• To introduce novel Gene Set Enrichment Analysis (GSEA) methods, gsGene and gsPG, specifically designed to overcome the limitations of analyzing DNA methylation data.
• To improve the accuracy, statistical power, and computational efficiency of GSEA for DNA methylation studies.
• To provide a robust framework for identifying biologically meaningful pathways from methylation data.

Main Methods:

• gsGene aggregates gene-level association signals, correcting for probe dependency and probe number bias.
• gsPG utilizes summary statistics from independent probe groups, mitigating biases from multi-mapping probes.
• A novel beta distribution fitting strategy was developed for improved P-value estimation, offering a computationally efficient alternative to permutation-based methods.

Main Results:

• Both gsGene and gsPG demonstrated superior performance compared to existing methods in comprehensive evaluations on two large datasets.
• The novel methods enhance statistical power and effectively control type I error rates in DNA methylation enrichment analysis.
• The beta distribution fitting strategy provides a computationally efficient and accurate alternative for P-value estimation.

Conclusions:

• The developed GSEA methods, gsGene and gsPG, offer significant improvements for analyzing DNA methylation data, addressing key challenges and enhancing biological insights.
• The R package dmGsea, implementing these methods, provides a user-friendly and efficient tool for researchers working with Illumina 450K, EPIC, and mouse methylation arrays.
• These advancements facilitate more reliable and computationally efficient pathway analysis in epigenetics research.