Fig. 1

Identification of microbial community profiles associated with treatment outcome in diet therapy. a Schematic showing clinical study design for identifying diet-responsive (DR) and non-diet-responsive (NDR) dogs. Antibiotics (Abtx) and prednisone (Pred) treatments are indicated. Abbreviated Canine Chronic Enteropathy Clinical Activity Index (CCECAI) scores were assessed at four different time points in b DR (n = 20) and c NDR (n = 9) animals. d Ternary plot of phylum-level OTUs from top 5 most abundant phyla among healthy (right), DR (left), and NDR (top) animals. Only OTUs with a maximum relative abundance > 0.001 and existing in at least 10% samples are shown. Bubble size represents the log2-transformed mean abundance of each OTU. Axes show reads accounted for by each OTU in each group of animals (DR, NDR, and healthy), as a percentage of total (sum) reads observed for a given OTU across all three groups. Arrows indicate the corresponding axis directions for each point. Relative abundance of e E. coli and g C. perfringens in animals with active disease (day 0) and healthy dogs. Spearman’s correlation between log10 abundance of f E. coli or h C. perfringens and CCECAI disease score. The shaded areas show the 95% confidence interval, as implemented in the geom_smooth function in the ggplot2 R package. i Differentially abundant OTUs between DR and NDR animals at day 0. Y-axis value represents the log2 fold change for DR versus NDR. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 using the two-sided Wilcoxon signed-rank test. Spearman’s correlations in f and h are significant (P < 0.05) with correlation coefficients of 0.216 and 0.227, respectively