HIMA2: high-dimensional mediation analysis and its application in epigenome-wide DNA methylation data

Area of Science:

• Biostatistics
• Genomics
• Environmental Health

Background:

• Mediation analysis is crucial for understanding environmental exposure effects on health.
• High-dimensional data necessitates advanced mediation analysis methods.
• Existing methods may struggle with the scale of modern biological data.

Purpose of the Study:

• To introduce HIMA2, an enhanced method for high-dimensional mediation analysis.
• To improve the identification of mediators in complex exposure-health outcome pathways.
• To provide a robust tool for large-scale biological and environmental studies.

Main Methods:

• HIMA2 utilizes sure independence screening (SIS) for dimension reduction.
• A de-biased Lasso procedure is employed for accurate parameter estimation.
• A multiple-testing procedure controls the false discovery rate (FDR) for hypothesis testing.

Main Results:

• HIMA2 effectively reduces mediator dimensions in high-dimensional settings.
• The method demonstrates reliable performance in simulation studies.
• HIMA2 successfully identified DNA methylation markers mediating smoking's effect on lung function in the CARDIA study.

Conclusions:

• HIMA2 offers a powerful approach for high-dimensional mediation analysis.
• The method enhances the ability to discover biological mediators.
• HIMA2 has practical applications in environmental health and genomics research.