A Network Approach to DNA Methylation Aging Clocks

Antón Carcedo, PhD Student, Department of Physics, Umeå University

Biological age, a critical indicator of organismal health, is commonly estimated using DNA methylation clocks. These clocks utilize DNA methylation, an epigenetic modification involving the attachment of methyl groups (CH_3) to cytosine-guanine dinucleotide sites (CpGs). Such modifications influence gene transcription levels, affecting the state of the organism. Existing methylation clocks rely on various black-box machine learning methods, such as Lasso Regression and ElasticNet, for CpG site selection, often lacking transparency regarding biological relevance. To address this, we introduce a systematic approach based on correlation networks among CpGs and their associations with chronological age. Our method defines a broader ensemble of potential clock configurations, enveloping previously established clocks. This selection approach not only clarifies the rationale behind existing clock configurations but also reveals numerous additional viable clock options. Our findings enable deeper biological interpretations by highlighting specific clusters and subsets of CpGs, facilitating mechanistic investigations into DNA methylation and its role in aging.