Metabolic diversity in human populations and correlation with genetic and ancestral geographic distances

Researchers found significant ethnicity-associated differences for 71% of metabolites, with acylcarnitines showing higher variability than amino acids. A developed metabolic distance measure showed a low positive correlation with genetic and ancient geographic distances between ancestral populations. Outlier pairs suggested influences from both genetics and environment (e.g., shared environment reducing metabolic distance between Black and White groups). Machine learning models could accurately distinguish individuals from genetically distant groups (e.g., AUC=0.96 for Black vs. Chinese) but not genetically similar ones (e.g., AUC=0.51 for Hispanic vs. Native American). Some metabolites informative for predicting ancestry were also biomarkers for inborn metabolic disorders. Limitations include reliance on predefined ethnicity categories, different sample sources for metabolic and genetic data, and simplified geographic ancestry models. The work implies that population-level metabolomics can reveal metabolic diversity at birth, with potential implications for improving genetic disease screening.