This project will support ERC Consolidator grant MeMoSA. 

Two workflows for 1. high-throughput  comprehensive 2D metabolomics and lipidomics and, 2. rapid clinically applicable ambient ionization metabotyping, will be developed. Molecular fingerprints of our unique deeply phenotyped pediatric cohorts (1.5k children) will be generated and advanced machine learning algorithms will be used to predict metabolite abundances based on their sources, i.e., diet, lifestyle,
anthropometrics, microbiome, drug intake, psychological factors, clinical markers, etc. A combination of in vitro digestions, in
vivo humanized mice, and in silico experiments with selected source variables will be designed to contribute to our understanding of
source-metabolite causality. These mechanistic insights will be used to build dedicated intervention trials in children with specific
source-dominated metabotypes.