The emergence and spread of antibiotic resistance have become emerging threats to human health. The human gut is a large reservoir for antibiotic resistance genes. The gut resistome may be influenced by many factors, but the consumption of antibiotics at both individual and country level should be one of the most significant factors. Previous studies have suggested that the gut resistome of different populations may vary, but lack quantitative characterization supported with relatively large datasets. In this study, we filled the gap by analyzing a large gut resistome dataset of 1,267 human gut samples of America, China, Denmark, and Spain. We built a stacking machine-learning model to determine whether the gut resistome can act as the sole feature to identify the nationality of an individual reliably. It turned out that the machine learning method could successfully identify American, Chinese, Danish, and Spanish populations with F1 score of 0.964, 0.987, 0.971, and 0.986, respectively. Our finding does highlight the significant differences in the composition of the gut resistome among different nationalities. Our study should be valuable for policy-makers to look into the influences of country-specific factors of the human gut resistome.
Keywords: Antibiotic resistance gene; Machine learning; Metagenomics; Resistome.