Direction-aware functional class scoring enrichment analysis of infinium DNA methylation data

Area of Science:

• Epigenetics
• Genomics
• Bioinformatics

Background:

• Infinium Methylation BeadChip arrays are widely used for epigenome-wide association studies (EWAS).
• Existing pathway analysis tools for methylation data often fail to retain crucial directional information for mechanistic insights.
• Functional class scoring (FCS) methods are valuable for pathway enrichment but require directional data retention.

Purpose of the Study:

• To evaluate candidate FCS methods that preserve directional information in methylation array data.
• To introduce and validate a novel method, termed LAM (Limma-Aggregated Mean), for pathway analysis of methylation data.
• To demonstrate the utility of LAM in identifying differential pathway methylation in diverse biological and disease contexts.

Main Methods:

• Evaluation of several FCS methods using simulation data.
• Implementation of LAM: mean aggregation of probe limma t-statistics by gene, followed by a rank-ANOVA enrichment test.
• Application of LAM to lung cancer datasets (paired tumor-normal), chronological aging, in vitro fertilization (IVF) conceived infants, and 19 disease states.
• Comparison with existing over-representation analysis methods.

Main Results:

• LAM demonstrated superior performance in simulations compared to existing methods.
• LAM exhibited higher sensitivity and robustness in real lung tumor-normal datasets.
• Analysis revealed novel associations between differential pathway methylation and chronological age, IVF conception, and various disease states.
• LAM successfully identified hundreds of novel differential pathway methylation associations.

Conclusions:

• LAM is a powerful and robust method for detecting differential pathway methylation, complementing existing analytical approaches.
• The method's ability to retain directional information enhances mechanistic understanding of genomic regulation.
• LAM provides a reproducible framework for pathway analysis of DNA methylation data.