A Novel Framework for the Design of Minimized Epigenetic Clocks Using the Analysis of DNA Methylation Heterogeneity.

Despite the significant progress made in the development of epigenetic age (eAge) clocks designed to estimate the various aspects of aging, currently available models, generated using large DNA methylation microarray datasets, still cannot fully address the issues of batch effects and technical variation. This hinders the use of the publicly available eAge clocks in routine laboratory practice, and it motivates the development of cost-effective, custom epigenetic clocks that are tailored to the given biological subjects and research methods. In this study, we analyzed the local DNA methylation of mesenchymal stem cell samples during culture expansion using high-throughput targeted bisulfite sequencing (BS-seq). Using the obtained data, we trained a minimized eAge model based on a Random Forest Regression with Leave-One-Out Cross-Validation, which determines cell passage with good performance (MAE 1.094 and R(2) 0.897) and which is comparable to previous solutions. Using the advantage of BS-seq to analyze consecutive CpGs methylation patterns, we demonstrated that combining the analysis of average DNA methylation levels with local methylation heterogeneity scores-thereby reflecting stochastic DNA methylation dynamics-can improve the quality of the epigenetic clock models. Therefore, we propose a research strategy for creating customized epigenetic clocks using targeted BS-seq and provide a mechanistic conceptualization of how information on longitudinal changes in DNA methylation patterns can potentially be used for the assessment of specific aging aspects.