Dataset 1 from 'Protein complexes in Bacillus subtilis by AI-assisted structural proteomics'

We generated a whole cell interaction network using crosslinking mass spectrometry. We crosslinked B. subtilis cells in suspension with the membrane permeable crosslinker DSSO, which can be cleaved in the mass spectrometer to aid identification of the crosslinking peptides (Kao et al. 2011; Kolbowski et al. 2021). To crudely simplify the proteome prior to tryptic digestion, we lysed the crosslinked cells and separated the soluble proteome (Dataset 1), and the cell debris (Dataset 2) (see Fig. S1a and methods). These two samples were digested, crosslinked peptides were enriched and fractionated by cation exchange and then all fractions were further separated by size exclusion chromatography. To further increase the depth of analysis, we also separated the soluble proteome further by size exclusion chromatography to produce 9 pools that were each analyzed separately (Dataset 3) (Fig. S1A and methods). A 2% protein-protein interaction false discovery rate (PPI-FDR) was imposed on each of the datasets and together 560 protein interactions are reported at a combined FDR of 2.5% (with a 5% residue pair-level FDR (Lenz et al. 2020)) (Supplementary Table 1). These 560 protein interactions are underpinned by 1268 unique residue pairs.