Synergistic analysis of genome-resolved metagenomics and language-based machine learning reveals hidden probiotic genomes in the Indian healthy human gut microbiome

Introduction The human gut microbiome plays an essential role in host physiology through metabolic activities such as micronutrient biosynthesis and maintenance of intestinal homeostasis. However, a substantial proportion of gut microorganisms remain uncultured, limiting the characterization of their functional roles and probiotic attributes. Genome-resolved metagenomics enables the recovery of microbial genomes directly from metagenomic data, facilitating the exploration of these uncultivated taxa. Methods Shotgun metagenomic datasets from healthy Indian individuals (n = 110) were analysed using the MuDoGer genome-resolved metagenomic workflow to reconstruct metagenome-assembled genomes (MAGs). MAGs were assessed according to MIMAG quality standards, taxonomically classified, and screened for probiotic characteristics using a machine learning-based prediction classifier. Biosynthetic pathways involved in B-complex vitamins and vitamin K production were identified through comparative genomic analysis. Results The analysis reconstructed 901 MAGs, including 289 high-quality genomes. Taxonomic classification identified 10 bacterial phyla and 109 genera, with Bacillota (47%) and Bacteroidota (41%) dominating the gut microbiome. Probiotic prediction identified 45 candidate probiotic genomes, comprising 22 culturable and 23 unculturable species. The unculturable species Megasphaera sp000417505 (15 MAGs) was the most abundant predicted probiotic taxon. Comparative functional analysis showed a higher abundance of B-vitamin biosynthesis genes in unculturable genomes, whereas culturable genomes contained more vitamin K biosynthesis genes. Discussion These findings indicate that both culturable and uncultured gut microorganisms contribute to probiotic-associated functions and micronutrient biosynthesis. Integrating genome-resolved metagenomics with machine learning provides a powerful framework for identifying candidate next-generation probiotics from complex microbial communities.