Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: The influence of the gut microbiome on BCG-induced trained immunity

Fig. 1

Study design overview and de novo assembly-based analysis of stool microbiome samples. A Study design overview. Healthy adult volunteers (N = 321) were vaccinated with BCG at baseline, after stool and blood samples were taken for metagenomic, metabolomic, and ex vivo cytokine production analyses. PBMCs isolated from subjects at baseline, after 2 weeks, and after 3 months were stimulated with S. aureus and M. tuberculosis ex vivo. B Association discovery strategy. Abundance of individual metagenomic species pangenomes (MSPs) was compared to fold changes in cytokine expression after 2 weeks and 3 months. MSPs significant in either modality were subsequently assessed for enrichment of associated circulating metabolites, while metagenomic assembly was searched for enzymes explaining the differential metabolomic potential. C Microbiome profiles were quantified as 345 MSP abundances, assembled with MSPminer software. MSPs were matched to known taxa on the phylum (all 345), genus (315), and species (208) levels. Barplot shows annotation levels of MSPs split in quartiles by number of genes. D Dendrogram of samples based on Jensen-Shannon distance between relative MSP abundances reveals three clusters. The cutoff point (dotted line) is the maximum mean difference in between-cluster and within-cluster distances (see Figure S1). E Multi-dimensional scaling (MDS) plots of samples’ microbial abundances based on Jensen-Shannon distance. Samples are colored by cluster (C1–C3, as in D) and cluster sizes are noted in parentheses. The three most abundant phyla (Firmicutes, Bacteroidetes, and Actinobacteria) are highlighted. F Per-sample relative abundances of detected phyla. Percentages in parentheses report cohort-wide abundances. G Number of MSPs significantly associated with questionnaire variables, stratified by phylum. Color coding as in F. The change in relative abundance of each MSP by each individual variable was assessed by Maaslin2 generalized linear model (FDR < 0.25)

Back to article page