Meta-analysis of human gut microbiome and blood metabolome associations

The blood metabolome contains both biomarkers and causative agents of health and disease. The microbiome is one of the most predictive features of the blood metabolome, yet the impact of specific microbiota and their functions on specific metabolites, as well as the specificity of this impact in cohorts from diverse ethnical, geographic and health backgrounds, remain unclear. To elucidate the relationship between microbiome and blood metabolome, we aim to conduct a meta-analysis of microbiome-metabolome associations using a machine learning approach. We will collect multiple studies of paired microbiome and metabolome measurements from diverse cohorts and processed this data in the same pipeline to make them comparable. The data requested here from Lifelines DEEP is one of these studies. To increase diversity, we also measured microbiome and metabolome in a  cohort of Japanese study participants with and without and intervention of a nutraceutical. We will with a panel of machine learning algorithms to predict each metabolite from each study, and used feature attribution analysis to identify the microbiota and pathways underlying each prediction.