IDDF2024-ABS-0039 Gut microbiota fingerprinting as a potential tool for tracing the population’s geographical origin

<h3>Background</h3> Human gut microbiota are individual specificity and temporal stability that have revealed significant compositional differences across geographical provenience. Previous studies have focused on comparing gut microbiota diversity among individuals across continents. However, the gut microbiota variations among people residing in different regions within a province remain enigmatic. <h3>Methods</h3> Shotgun metagenomics sequencing was performed to analyze the gut microbiota of 381 unrelated Chinese Han individuals living in high-income city and traditional city of Hubei province. Propensity score matching was implemented to mitigate potential biases. The difference between two distinct regions was investigated using machine learning to identify the discriminatory ability of the gut microbiota. <h3>Results</h3> A total of 77 high-income city and 108 traditional city individuals were matched after propensity score matching. No significant differences were observed in the microbial α-diversity and β-diversity. The gut microbiota of high-income city individuals exhibited a higher relative abundance of <i>Blautia</i> genus. Conversely, the microbiota of traditional city people demonstrated a higher relative abundance of <i>Lachnospira</i> genus. Additionally, <i>Roseburia faecis</i>,<i> Lachnospira pectinoschiza</i>, <i>Flavonifractor plautii</i>, and other 9 species were found to be significantly different between the two regions. Furthermore, three prediction models based on the random forest, support vector machine, and logistic regression algorithms were constructed. Of the test samples, 86.1% could be classified with the random forest model based on 85 species, achieving an area under the receiver operating curve (AUC) of 0.895 (95% CI, 0.784-1.000). <h3>Conclusions</h3> The gut microbiota of individuals residing in the same province exhibits significant similarity, however, pronounced differences in bacterial assemblages were noted between individuals from high-income cities and traditional cities. We hypothesize that leveraging the machine learning algorithms to enhance the discrimination between two regional populations’ microbiota can facilitate a deeper understanding of host-specific associations, which could offer valuable clinical assistance in diagnosis and treatment.