Fig. 2: Validation and characterization of structural variations (SVs) in human gut microbiome.

a Schematic representation of direct validations of structural variations (SVs) using long ONT reads. Using the upper metagenome-assembled genome (MAG) as reference, deletions (left) and insertions (right) were identified in the lower MAG (belonging to the same bacterial species, from a different sample), and mapping long reads from different sample against representative sequences resulted in reads directly covering deletion (left) or insertion (right) and flanking regions, thus validating the presence of these SVs at read-level. b, c Phylum-level and family-level distributions in the number of major types of SVs (insertions and deletions) across different taxonomical groups, corrected for corresponding genome size (SVs per 1 M genome). High variability of SV numbers can be found among different phyla and bacterial families (see Results). d Comparison of average number of SVs per 1 Mb genome between all of 189 MAGs used for SV detections, as well as individual MAGs from the three most common same species, between different individuals in the cross-sectional cohort (left of each boxplot) and from different samples within the same individual in the time-series data (right of each boxplot). In all four cases, inter-individual SVs numbers are significantly higher than that of intra-individuals (two-sided Wilcoxon test, n = 4093, 83, 91, 63 separately, all P < 2e-10), suggesting SVs can be used as fingerprints in human gut microbiome to distinguish different individuals. Data are presented as box plots with whiskers at the 5th and 95th percentiles, the central line at the 50th percentile, and the ends of the box at the 25th and 75th percentiles.