Fig. 1
Cheek microbiomes were differentiated into cutotypes and associated with PAH exposure. a Principal coordinate analysis (PCoA) plot based on the Bray–Curtis dissimilarity of microbial community composition. Data points represent metagenomic samples with point size representing Shannon diversity and color showing the cutotype as determined by the prediction strength analysis. The inset panel shows the pairwise Bray–Curtis dissimilarity between samples of the same or different cutotypes, showing that microbiomes within a cutotype were more similar than between cutotypes (**Tukey’s pairwise post-hoc p < 0.05 for all pairwise comparisons). b PCoA plot as shown in panel a with the data points colored according to the relative abundance of Cutibacterium acnes, Xanthomonas citri, and Rhodococcus opacus in cutotypes 1 and 2. c Top 12 species-level bacterial taxa based on the average relative abundance in all the samples organized by cutotype. Other bacterial taxa were grouped into the “Minor/Unclassified” category. d Distance-based redundancy analysis depicting the effects of exposure to high and low molecular weight PAHs (purple and green, respectively) on cutotype differentiation. PAHs with a molecular weight < 200 g/mol were considered low. e Spearman’s correlation between the first component of the PCoA and the measured concentrations of the six PAHs. Linear regression lines and R2 values are shown. f Association of samples in the respective cutotypes with different host factors and host skin phenotype