Fig. 2.

Key members of DD cutaneous microbiome correlate with disease severity and may serve as microbial disease markers. A Correlation analysis of key taxa with α-diversity (richness and Shannon index), DD score (global severity score) and ODD score (objective severity score). B Linear discriminant analysis effect size (LEFSe) of key taxa distribution over the different skin locations. LEFSe employs the Kruskal–Wallis rank sum test to detect OTUs with significant differential abundances between groups (control, NIDS and IDS), the pairwise Wilcoxon test between sub-groups, followed by a linear discriminant analysis (LDA) to evaluate the relevance or effect size of each differentially abundant taxon. The heatmap to the right shows in which group the taxa relative abundance is increased and the LDA score displays the potential marker taxa of the group. We considered in our analysis that the strongest associations between key taxa and skin phenotype are best delineated above the LDA cutoff of 5.5. Red and blue colors on the LDA plot indicate high and low taxa relative abundances, respectively. C Correlation network of microbiome communities in IDS and control groups displayed at the genus level. The SparCC (sparse correlations for compositional data) approach has been used to define network associations. This approach assumes a sparse network and performs iterations to identify taxa pairs that are outliers to background correlations. Each node represents a taxon and its size is proportional to the number of connections. The green and orange colors on the nodes indicate the taxa mean relative abundance in the control and IDS groups, respectively. Taxa are only connected if the correlation meets a p value cutoff of 0.05 and a correlation coefficient of 0.3. Key correlations with the Staphylococcus genus are highlighted in the figure with blue and red lines, representing negative and positive correlations, respectively. D Correlation analysis of key taxa relevant in DD pathology. E Linear correlation plots of representative taxa interactions. Pearson’s coefficient was used to calculate correlations among taxa or between taxa and metavariables. Red and blue colours on the correlation plots respectively indicate positive and negative correlations. Statistical significance was calculated using Kruskal–Wallis and Wilcoxon-Mann–Whitney tests, respectively, for multiple group and pairwise comparisons. The asterisks indicate statistically significant differences and correspond to *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001