Figure 2.

Integrating molecular and association networks. As the size of the network is enormous (41,765 nodes and 45,479 edges), here, we focus on some of the networking families that have a known molecule identified by Dereplicator+ [34] as a polyphenol or a vitamin (25 molecules in total). The network for all 25 molecules is shown in Figure S5. As the association network is currently too dense to visualize, we only show the top two microbial features for each molecular feature (Fisher p-value of ${10}^{-5}$). The known transformations were reported by BioTransformer [29]. By focusing solely on edges in the molecular network and association network, one can get a large number of potential biotransformations with a high chance of being spurious. However, focusing on the overlap between the two networks (golden triangles) results in a much smaller set of potential biotransformations, where many of them can be validated by a literature search.