Abstract
Age is a major risk factor for various diseases, such as cancer, cardiovascular conditions, and neurodegenerative diseases. However, chronological age, the simple number of years one has lived, does not capture individual health differences, prompting the development of methods to accurately estimate biological age instead of relying on chronological age. One of the major molecular approaches exploits DNA methylation (DNAm), which is an essential epigenetic modifier for regulating gene expression, cell differentiation, and aging. DNAm-based aging clocks have been developed to predict biological age, but the prediction is highly dependent on training data, including organs and assay technologies. To address these clocks' high variance, we present two EnsembleAge clocks, leveraging eight previously developed DNAm clocks, harnessing the strengths of each methylome age clock, smoothing out individual variances, and providing a more robust estimation of biological age. We trained our EnsembleAge clock models using DNA methylation data from nine organs in the Genotype-Tissue Expression (GTEx) dataset. Our EnsembleNaive clock model achieved the lowest median absolute error (MeAE) of 4.04 years in whole blood. The EnsembleLR model demonstrated the lowest MeAE of 6.35 years across multiple tissues, including breast, lung, muscle, ovary, prostate, testis, and colon. The significant reduction in MeAE underscores its high practical value in clinical and forensic applications, especially in contexts where epigenetic changes are subtle. We further applied our models to four public datasets representing diverse biological applications, including administered short-term medical opioid use (GSE151485) and long-term opioid overdose (GSE164822). Our model reveals over 10 years of age acceleration in opioid-overdosed brains but no significant epigenetic age acceleration when opioid usage was well administered. Our EnsembleAge clock models are also implemented as a web service, allowing users to conveniently upload their DNA methylation data and receive predictions of their biological age. This empowers individuals to track their biological/epigenetic age over time, mitigating the effect of variance and promoting healthy aging and a beneficial lifestyle. Our EnsembleAge clock service is available at https://ensemble.epiclock.app/ .