Spectral Networks for Peptide Identification

The concept of spectral networks^{37 (link)} (also known as molecular networks^{44 (link)} in the field of natural products) was introduced to reveal spectra of related peptides within a proteomic dataset without knowing what these peptides are. While these networks were first introduced for linear peptides, they were later generalized to cyclic peptides and metabolites^{44 (link),61 (link)}. Nodes in a molecular network correspond to spectra, while edges connect spectra that are generated from related metabolites (e.g., metabolites differing by a single variation). The variations that are captured by molecular networks help to infer mutations, modifications (such as oxidation and acetylation), or adducts (such as sodium and potassium).
DEREPLICATOR+ constructs the molecular network of all spectra and selects connected components with at least one identified spectrum. Using the MSM corresponding to this spectrum, it annotates variants of identified metabolites. Supplementary Figure 10 shows the molecular network of the chalcolmycin family identified by DEREPLICATOR+.