Abstract
The interplay between host epigenetics and microbial metabolites is a dynamic interface affecting human health. Microbial small molecules — short-chain fatty acids, bile acids, and tryptophan metabolites — influence host epigenetic mechanisms, including histone modification and DNA methylation, with downstream effects on gene expression. Predicting these interactions remains challenging due to complexity, context dependence, and high dimensionality of biological data. Here, we propose a reinforcement learning (RL) framework to model host-microbe cross-talk with the goal of forecasting epigenetic changes induced by microbial metabolites. We integrate multi-omic datasets (microbiome taxonomic/functional profiles, metabolomics, host epigenetic marks) into a reward-based learning system to enable predictive modeling, policy optimization, and hypothesis generation.Therapeutic modulation of host epigenetic states via targeted microbiome alteration.
