A0423
Title: Phenotype-free statistical model for DNA methylation-based epigenetic clocks via reflective CpG site selection
Authors: Bo-Han Yao - National Cheng Kung University (Taiwan) [presenting]
An-Shun Tai - National Tsing Hua University (Taiwan)
Abstract: DNA methylation-based epigenetic clocks, particularly second-generation models incorporating phenotypic data, have advanced biological age prediction. However, most existing clocks are developed within specific populations and conditions, limiting their applicability to other contexts. As a result, constructing population-specific clocks is often recommended. Nonetheless, phenotype adjustment required by current epigenetic clock construction methods may be infeasible in settings with limited data availability. To address this issue, the novel concept of reflective CpG sites is introduced, which are primarily associated with exposure-related aging fluctuations, and a mixture model-based approach is proposed to identify these sites for constructing epigenetic clocks without relying on phenotype data. The performance and broad applicability of the proposed method are demonstrated through simulation studies.
