Despite extensive efforts to address it, the vastness of uncharacterized ‘dark matter’ microbial genetic diversity can impact short-read sequencing based metagenomic studies. Population-specific biases in genomic reference databases can further compound this problem. Leveraging advances in hybrid assembly (using short and long reads) and Hi-C technologies in a cross-sectional survey, we deeply characterized 109 gut microbiomes from three ethnicities in Singapore to comprehensively reconstruct 4497 medium and high-quality metagenome assembled genomes, 1708 of which were missing in short-read only analysis and with >28× N50 improvement. Species-level clustering identified 70 (>10% of total) novel gut species out of 685, improved reference genomes for 363 species (53% of total), and discovered 3413 strains unique to these populations. Among the top 10 most abundant gut bacteria in our study, one of the species and >80% of strains were unrepresented in existing databases. Annotation of biosynthetic gene clusters (BGCs) uncovered more than 27,000 BGCs with a large fraction (36–88%) unrepresented in current databases, and with several unique clusters predicted to produce bacteriocins that could significantly alter microbiome community structure. These results reveal significant uncharacterized gut microbial diversity in Southeast Asian populations and highlight the utility of hybrid metagenomic references for bioprospecting and disease-focused studies.
Reference genomes for gut microbiomes help unravel microbial “dark matter” and serve as valuable resource for disease-focused studies. Here, the authors perform short and long read metagenomics and metagenome-assembled genomes analyses to profile the gut microbiome of Southeast Asian populations, revealing significant species and strain-level diversity, with thousands of previously uncharacterized biosynthetic gene clusters.