R methylCIPHER: A Methylation Clock Investigational Package for Hypothesis-Driven Evaluation & Research

Abstract

Background: Epigenetic clocks are promising tools for the study of aging in humans. The clocks quantify biological aging above and beyond chronological age, demonstrate systematic associations with risk factors that accelerate aging, and predict age-related morbidity and mortality. There is interest in using them as surrogate endpoints in intervention studies. However, the large number of clocks, decentralized publication and explosive popularity in the last decade has made for poor accessibility and standardization. This has hampered the abilities of new researchers to conduct truly hypothesis driven research--whether by not knowing about the best available clocks for a given question, or by systematically testing many or all as they become available. Results: We report a centralized R package which can be installed and run locally on the user's machine, and provides a standardized syntax for epigenetic clock calculation. The package includes a set of helper functions to assist with navigating clock literature and selecting clocks for analysis, as well as affording the user with the details of clock calculation. We describe each clock's resilience to missing CpG information, combined with functionality to assess the need for imputation in the user's own data. Furthermore, we demonstrate that while CpGs may not be shared among clocks with similar outputs, many clocks have highly correlated outputs. Conclusions: Due to the previous decentralization of epigenetic clocks, gathering code and performing systematic analysis, particularly in protected datasets, has required significant information gathering effort. Here, we offer an R package with standardized implementation and potential for future growth and clock incorporation to assist with hypothesis driven investigation of aging as measured by epigenetic clocks. We show the potential of this package to drive the user to think globally about signals captured by epigenetic clocks, as well as to properly identify the potential and limitations of each clock in their current research.

Refining epigenetic prediction of chronological and biological age

“…Our final cAge predictor (trained on all 11 datasets in Table 1) and those by Zhang et al (ZhangAge) [5], Hannum et al (HannumAge) [4], and Horvath (HorvathAge) [3] were projected onto the GSE55763 dataset to compare their performance in an independent test set. External clock predictions were calculated using the methylCI-PHER R package [38] (https:// github. com/ Morga nLevi neLab/ methy lCIPH ER).…”

A Unified Framework for Systematic Curation and Evaluation of Aging Biomarkers

“…The modular design of Biolearn encourages the addition of new models and datasets, making it a living library that will grow in tandem with the field itself. By centralizing resources and knowledge, Biolearn considerably reduces redundancy and accelerates biomarker development and validation efforts Previous efforts to harmonize biomarkers of aging, notably methylCIPHER and BioAge 25,26 , have been limited in scope, focusing on methylation or blood-based biomarkers only.…”

The Sociodemographic and Lifestyle Correlates of Epigenetic Aging in a Nationally Representative U.S. Study of Younger Adults

“…Beta values for CpG sites across the genome were measured according to kit protocols and filtered to remove polymorphic positions. Remaining CpG sites were restricted to a set of 30,484 CpG sites used with the DNA methylation calculator (https://dnamage.genetics.ucla.edu) and Methylcipher 44 and Dunedin 33,34 1; details in Supplement 2). Clocks are shown in order of their "generation" beginning with five first-generation clocks: Horvath1, Horvath2, Hannum, VidalBralo, and Zhang2019.…”

“…7,1413,40 Some recent studies have used large representative samples that examined a range of clocks, but focused on older adults. 10,41 In fact, the majority of research on epigenetic aging has been based on samples of older adults 10,1415,20,[22][23][24]38,[42][43][44][45] (exceptions include 46,47 ). Given that older adults are more likely to have chronic comorbidities, it is difficult to disentangle the impacts of underlying disease from sociodemographic exposures.…”