Fig. 1: The pipeline for RiPP discovery by HypoRiPPAtlas.

a The pipeline extracts BGCs from microbial genomes available at RefSeq and IMG/M¹¹, or uses the readily available BGCs from IMG-ABC¹¹, antiSMASH database¹², and BiG-SLiCE¹³. Colored boxes represent annotated domains. b HypoRiPPAtlas is constructed by predicting the hypothetical molecule structures from the BGCs. Three hypothetical structures are distinguished and labeled using different colors. c The Atlas is further filtered down to specific taxonomies/gene clusters based on the taxa/metagenomic information available from the samples of interest. d Mass spectrometry fragmentation of the hypothetical molecules in the Atlas are predicted, along with known RiPPs from PubChem⁷⁴ and NP-atlas⁷⁵. e Mass spectra are collected on the environmental samples/microbial isolates, e.g. from a GNPS dataset¹⁵. f Mass spectra are searched against the predicted spectra of hypothetical molecules, and high-scoring RiPP-spectrum matches are reported. g The identifications are further expanded through propagation in the molecular network¹⁵. Steps (a–d) are done only once and stored in a repository, while steps (e–g) are repeated for every new mass spectral dataset.