The Epilysian Genome: AI-Driven Insights into Healthy Epigenetic Aging
Description: Aging is a multifaceted process influenced by genetic, environmental, and epigenetic factors. While genetics explain only part of lifespan variability, epigenetic modifications critically shape biological aging and disease susceptibility. Centenarians offer a unique opportunity to explore protective epigenetic signatures behind exceptional longevity. This project aims to develop a transformer-based machine learning model that integrates DNA methylation data with 3D genome architecture to improve biological age prediction and pinpoint key epigenetic determinants of longevity. By constructing a 3D Informer Matrix from Hi-C chromatin interaction data, we capture spatial dependencies between CpG sites, overcoming the limitations of traditional linear models and leverage attention-based deep learning to decipher complex aging patterns. We will (1) construct the 3D Informer Matrix, (2) integrate it into a transformer-based methylation model, (3) analyze attention mechanisms to pinpoint epigenetic signatures distinguishing centenarians and (4) perform in silico counterfactual analyses on methylation changes. Comparing centenarians with progeria patients will further reveal regulatory loci linked to resilience or susceptibility to aging. This integration of 3D genomic structure with epigenetic profiling is a complete novel aproach which aims to refine age prediction, deepens our understanding of aging mechanisms, and identifies potential intervention targets to extend healthspan.
Funder: Fonds Wetenschappelijk Onderzoek
Funding type: fellowship
Funder scheme: FWO fellowships
Start date: 2025-11-01
End date: 2029-10-31
Investigator: Niel Vanhamel
Institution: UGent
Funded outputs count: 0
Funder award ID: 11A2O26N
OpenAlex ID: G4488130146