Enhancing the performance and interpretability of epigenetic clocks.

Area of Science:

• Epigenetics
• Genomics
• Aging Research

Background:

• Epigenetic clocks using DNA methylation (DNAm) accurately predict chronological age.
• The precise biological mechanisms underlying epigenetic clock accuracy, particularly the role of DNAm in gene regulation, are not fully understood.
• Transcription factor (TF) binding activity is a key mechanism of gene regulation potentially influenced by DNA methylation.

Purpose of the Study:

• To investigate the regulatory potential of CpGs utilized in established epigenetic clocks.
• To determine if DNA methylation changes at transcription factor binding sites (TFBS) contribute to the accuracy of epigenetic clocks.
• To develop an improved epigenetic clock model by integrating regulatory information.

Main Methods:

• Analysis of CpGs in established epigenetic clocks for overlap with known transcription factor binding sites (TFBS).
• Identification of transcription factors associated with age-related DNA methylation changes.
• Development of a novel TFMethyl Clock model using TFBS-associated CpGs and feature engineering.

Main Results:

• Most CpGs in current epigenetic clocks do not overlap with TFBS, suggesting clock accuracy isn't solely driven by TF binding dynamics.
• Specific TFs like ZBED1, NFE2, and CEBPB were enriched for age-associated CpGs, while RELA, IKZF1, and STAT3 showed protective effects.
• The TFMethyl Clock model achieved competitive age prediction accuracy and identified target genes involved in inflammation and metabolism with significant age-related DNA methylation changes.

Conclusions:

• Epigenetic clock accuracy is not primarily driven by DNA methylation changes at TFBS.
• Incorporating regulatory information, such as TFBS data, into epigenetic clock models can offer mechanistic insights into aging.
• The TFMethyl Clock model demonstrates the potential for improved interpretability and predictive power by considering regulatory elements.