Weighted 2D-kernel density estimations provide a new probabilistic measure for epigenetic age

Area of Science:

• Gerontology
• Epigenetics
• Biostatistics

Background:

• Epigenetic aging signatures offer insights into human aging.
• Traditional epigenetic clocks often use linear regression, assuming linear DNA methylation trajectories.
• Non-linear epigenetic aging patterns may provide a more comprehensive understanding of biological age.

Purpose of the Study:

• To explore a non-parametric approach for epigenetic age determination using 2D-kernel density estimation.
• To develop a weighted model for improved epigenetic age prediction.
• To introduce a variation score for quantifying age-related epigenetic variability.

Main Methods:

• Utilized 2D-kernel density estimation for epigenetic age calculation.
• Developed a weighted model based on this non-parametric approach.
• Assessed predictive accuracy against conventional epigenetic clocks.
• Analyzed a variation score derived from the model in a human cohort.

Main Results:

• The weighted 2D-kernel density estimation model achieved predictive accuracy comparable to conventional epigenetic clocks.
• A novel variation score was generated, reflecting inherent variability in age-related epigenetic changes.
• This variation score was significantly elevated in various disease states.
• The variation score demonstrated a significant association with mortality risk in the Lothian Birth Cohort 1921.

Conclusions:

• Weighted 2D-kernel density estimation provides an accurate method for epigenetic age prediction.
• The derived variation score serves as a valuable additional variable for biological age estimation.
• This non-parametric approach enhances our understanding of epigenetic aging dynamics and disease association.