Fig. 2

The enterotypes of the human gut mycobiome. a Clustering results of fungal enterotypes on ITS1 and ITS2 datasets and visualized by principal coordinate analysis (PCoA), and the most abundant genera within each enterotype is shown. The between-sample distances within each cluster compared to the median distance between clusters (black line) are shown at the bottom right of each panel. The bar height is the median distance, and the whiskers represent the 25th and 75th quantiles. b A four-enterotype classifier trained on the ITS2-sequencing datasets was applied to predict enterotypes in the ITS1-sequencing datasets, and the corresponding Area Under the Receiver Operating Characteristic Curve (AUC) values were presented. “Without drivers” refers to excluding the driver genera Candida, Saccharomyces, Aspergillus, Saccharomycetales sp. and Ascomycota sp. when training the classifiers. c The concordance of enterotype-associated fungal genera and enrichment trends across different cohorts, and log(FC) denotes the log-transformed fold change of the average relative abundance of the genera within respective enterotypes relative to that of others. The taxa name with a placeholder means that it could not be confidently assigned to a known taxonomic group. Asterisks represent the statistical significance of the multiple testing corrected one-sided Wilcoxon rank-sum tests (*adjusted p < 0.05, **adjusted p < 0.01, ***adjusted p < 0.001). d The correlations between fungal enterotypes and bacterial enterotypes in the CHGM cohort. The color reflects the O/E ratio (the ratio of observed count to expected count), and asterisks represent the statistical significance of Fisher’s exact test for each pair of comparison: *p < 0.05, **p < 0.01