Ordering molecular diversity in untargeted metabolomics via molecular community networking

Molecules in living systems are not random but are shaped by biological necessity. Mass spectrometry (MS) is a powerful tool for exploring these complex molecular landscapes. Molecular networking links metabolites by spectral similarity, but conventional methods leave many nodes disconnected. We introduce molecular community networking (MCN), which identifies natural molecular clusters and prunes them to keep the strongest links. The approach increases connectivity to about 95% of molecules and better captures structurally related compounds, including distinct ion forms and in-source fragmentation ions. MCN also improves the mapping of molecular space, helping distinguish true novel molecules from artifacts. Using MCN, we discovered dipeptide-conjugated bile acids associated with Bifidobacterium breve and proposed structures for previously unexplored N-acyl amides that interact with G protein-coupled receptors. We also built a global metabolome map from public GNPS/MassIVE data, covering about 8.4 million molecular features, creating a "roadmap" for molecular diversity.