Utilizing metagenomic profiling and machine learning model to identify bacterial biomarkers for major depressive disorder.

Background: Major depressive disorder (MDD) is highly heterogeneous, which provides a significant challenge in the management of this disorder. However, the pathogenesis of major depressive disorder is not fully understood. Studies have shown that depression is highly correlated with gut flora. The objective of this study was to explore the potential of microbial biomarkers in the diagnosis of major depressive disorder.

Methods: In this study, we used a metagenomic approach to analyze the composition and differences of gut bacterial communities in 36 patients with major depressive disorder and 36 healthy individuals. We then applied a Support Vector Machine Recursive Feature Elimination (SVM-RFE) machine learning model to find potential microbial markers.

Results: Our results showed that the alpha diversity of the intestinal flora did not differ significantly in major depressive disorder compared to healthy populations. However, the beta diversity was significantly altered. Machine learning identified 8 MDD-specific bacterial biomarkers, with Alistipes, Dysosmobacter, Actinomyces, Ruthenibacterium, and Thomasclavelia being significantly enriched, while Faecalibacterium, Pseudobutyrivibrio, and Roseburia were significantly reduced, demonstrating superior diagnostic accuracy (area under the curve, AUC = 0.919). In addition, the gut bacteria performed satisfactorily in the validation cohort with an AUC of 0.800 (95% CI: 0.6334-0.9143).

Conclusion: This study reveals the complex relationship between gut microbiota and major depressive disorder and provides a scientific basis for the development of a microbiota-based diagnostic tool for depression.