RelAge-GNN: Graph Neural Network for DNA Methylation Age Prediction
RelAge-GNN, a novel machine learning framework, leverages multi-relational graph neural networks to assess biological age based on DNA methylation data. In contrast to traditional techniques that analyze CpG sites in isolation, RelAge-GNN develops three interrelated graphs that represent co-methylation patterns, genomic co-localization, and gene-level connections among CpG sites. Each graph is handled by a separate GNN branch, with a learnable gating mechanism that dynamically integrates the representations. This methodology seeks to enhance predictive accuracy by capturing intricate biological interactions. The research can be found on arXiv.
Key facts
- RelAge-GNN is a multi-relational graph neural network framework for DNA methylation-based age prediction.
- It constructs three complementary graphs: co-methylation patterns, genomic co-localization, and gene-level associations.
- Each graph is modeled by an independent GNN branch.
- A learnable gating mechanism adaptively fuses the representations.
- Most existing methods treat CpG sites as independent features.
- DNA methylation is a stable biomarker strongly associated with aging.
- Aging clocks estimate biological age from observable biomarkers.
- The paper is published on arXiv with ID 2605.07175.
Entities
Institutions
- arXiv