Fig. 1

Merged host/microbe metabolic network. a Each node (circle) is a reaction in the host genome (left) or microbial metagenomes (right). Host nodes are colored purple if derived from an orthology association between sheep and humans, tan if from an ovine/bovine relationship or blue for the case of the added buyrate-employing pseudo-reaction (Materials and Methods). Edges are shared metabolites (network N ₅₀ ). In the center are nodes employing 23 potentially shared compounds between the host and microbes (set VFA + AA; Materials and Methods): the 10 most frequent metabolites (by microbe read count to their respective reactions) are individually colored. All nodes are organized by their distance from the other subnetwork: hence nodes employing an interface metabolite are at the center with a distance of 0. Microbial nodes are colored based on the normalized log₂-fold difference in read count between the two diets (green: overabundant in the FORG diet; red: overabundant in CONC). Nodes whose normalized read counts did not differ significantly between the diets are shown in black (Materials and Methods). b The right half of part a, recolored based on the normalized animal-to-animal variance in read count for the FORG animals (see scale bar). c Same as for b but for the CONC animals. d Histogram giving the cumulative proportion of the total mapped reads for the two diets (green: FORG; red: CONC) at each level of the network. Note that the FORG animals have proportionally more reads mapped to more distant layers of the network