A multi-omics analysis of gut bacteriome, virome, and serum metabolome in bipolar depression

The involvement of microbiota-gut-brain axis in bipolar disorder (BD) has been uncovered, yet the specific tripartite interplay between the gut bacteriome, virome, and serum metabolome remains to be elucidated. We conducted a cross-sectional multi-omics analysis on 90 drug-free patients with bipolar depression and 30 healthy controls. A significant between-group difference in gut bacterial α-diversity was observed. Non-parametric test revealed 1929 bacterial and 134 viral species with significant inter-group difference, among which 249 bacterial and 7 viral species remained significant after FDR correction (P_adjusted < 0.05). Metabolomic analysis identified 261 significantly differential serum metabolites, which were enriched in 70 biological pathways and 40 pathways remained significant after correction. Integration of the datasets revealed strong cross-omic correlations, while only eight significant viral-metabolic correlations were detected. Post-FDR significant correlations with clinical features were exclusively observed between differential metabolites and scores of disease severity, with a predominance of negative correlations. Clinically, a random forest model integrating bacteriome, virome, and metabolome features achieved superior discriminative power (AUC = 0.986) compared to single-omics models (metabolites: 0.970; bacteria: 0.823; viruses: 0.732). This work demonstrated a dysregulated bacteriome-virome-metabolome network of patients with bipolar depression, providing a robust panel of candidate biomarkers for the precise diagnosis of BD.