Figure 6.

Altered proportions of gut bacterial, archaeal, fungal, viral, and parasitic communities in COVID patients are associated with disease severity. A. Distribution of proportional read (log2 scale) representation of five different microbial kingdoms in the shotgun metagenomes as a function of known disease severity (n = 718); *p < .05, **p < .01 (Fatal versus other disease category, Wilcoxon rank-sum test). B. COVID disease severity is significantly associated with differences in the bacterial, archaeal, fungal, viral, and parasitic community proportions across cohorts (first stool sample only from each subject, n = 537); microbiome variation was measured by PERMANOVA based on Bray-Curtis dissimilarity matrices computed on relative abundance with cohort adjusted and marginal sums of squares applied, i.e., adonis2(relative abundance matrix ~ severity + cohort, permutations=999, method=“bray”, by=“margin”). Non-metric multidimensional scaling (NMDS) of microbiome-severity correlation based on species-level Bray-Curtis dissimilarity of pooled shotgun metagenomes with known disease severity (n =718) for (c) bacteria, (d) archaea, (e) fungi, (f) viruses, and (g) parasites; colored boxplots on the top and the right represent Bray-Curtis dissimilarity by disease severity in the first and second ordinations, respectively.