Journal of Molecular Evolution · Nov 13, 2026

A Structurally Interpretable ElasticNet-Based Epigenetic Clock Reveals Distributed and Individualized Architecture of Human Biological Aging

By Akrami M, Akrami M, Amini A et al.

DNA methylation patterns provide a remarkably accurate molecular readout of human biological aging. In this study, we leverage an ElasticNet regularization framework not primarily to optimize predictive accuracy, but to interrogate the structural architecture of the aging methylome characterizing its sparsity regime, per-CpG kinetic heterogeneity, and inter-locus correlation topology across the adult lifespan. The model achieved robust predictive performance, explaining 71% of chronological age variance (R² = 0.71) with a mean absolute error (MAE) of 6.87 years on independent test samples. Parameter optimization across regularization strengths revealed that the aging signal is moderately sparse distributed across hundreds rather than thousands of CpG sites indicating a non-trivial balance between redundancy and biological independence among age-associated loci. Dimensionality reduction via principal component analysis (PCA) uncovered a smooth and continuous age gradient embedded in the methylome, with younger individuals forming compact clusters and older individuals displaying increasing dispersion. This structure was faithfully reproduced when projecting test samples into the same space, underscoring the generalizability of the model. CpG-specific trend analysis revealed heterogeneous kinetic behaviors, from linear hypomethylation to stochastic variance, collectively capturing both deterministic and individualized aspects of epigenetic aging. Correlation mapping and graph-theoretic community detection among the top 50 CpGs demonstrated a statistically significant, sparsely modular organization (Newman’s modularity Q = 0.41, p < 0.001 vs. randomized null networks) rather than a singular coordinated program. Together, these findings describe an interpretable and biologically grounded methylation clock that not only predicts chronological age but also reflects inter-individual differences in epigenetic integrity. The residual deviations between predicted and true age may thus represent personalized trajectories of molecular aging quantifying the subtle interplay between time, environment, and individuality encoded in the human methylome.

doi.org/10.1007/s00239-026-10365-wPubMed 43024061Biomarkers

AgingResearch.News