This repository contains the R code of the algorithm Multi-Mean GPs, presented in the paper:

Longitudinal prediction of DNA methylation to forecast epigenetic outcomes
Leroy, A., Teh, A.L. et al.
eBioMedicine, Volume 115, 105709
https://doi.org/10.1016/j.ebiom.2025.105709

If you are considering citing this work, you are welcome to use the file reference.bib.

The 'Data' folder contains the coefficients of the PedBE and Horvath clocks used to compute epigenetic age in the paper. The data from the GUSTO project presented in the experiments can be downloaded on the dedicated online platform: https://gustodatavault.sg/.

The 'Training' folder contains the trained models for the different settings presented in the paper.

The 'Prediction' folder contains the prediction results presented in the paper.

The 'Evaluation' folder contains the evaluation metrics presented in the paper.

The 'MMGP.R' file contains the code of the Multi-mean GPs algorithm and a few convenient functions to conduct the experiments and evaluate and display the results.

The 'Experiments.R' file contains the scripts used to conduct the experiments and plot the graphs presented in the paper.

Corresponding author and primary contributor to the code: Arthur Leroy arthur.leroy.pro@gmail.com'