Fig. 9: An illustration of finding motifs through subgraph isomorphism and motif-based modification.

In (a) and (b), the query is highlighted in orange, while the subject is highlighted in blue. a First, the hydrogens are removed from the query and the subject. Next, the atoms are labeled by a breadth-first traversal of each graph from the first non-hydrogen atoms in the input file. Finally, an adjacency matrix for each graph is prepared. (b) Only the mappings that define an isomorphism are kept. Converting known enzymatic modifications of RiPPs into a computer-readable format for (c) dimethylation of N-terminal, e.g. in cypemycin⁷⁶ and (d) oxidative decarboxylation, e.g. cypemycin and epidermin⁷⁷. In our format, we use commands disconnect/connect (for bonds) and add/remove (for chemical substructures). For example, in part (c), “disconnect 1 5” removes the bond between the nitrogen atom with index 1 and the hydrogen atom with index 5, while “remove 5” removes the hydrogen atom with index 5, and “Add 1 CH₃” adds a methyl group to the nitrogen atom indexed 1 (methylation). In part (d), “connect 9 14 2” adds a double bond between atom 9 and atom 14. Each step of the modifications is differentiated by different colors, both in the action on the right and in the molecule on the left. We converted all the modifications from Arnison et al.⁴, Montalbán-López et al.⁷⁸, and our in-house database of modifications into this format (Supplementary Figure 24 and Supplementary Table 6).