Fig. 3: Protein co-expression clusters determined by neural network-based clustering allow improved annotation.

a Principal component plot of the 1444 protein groups quantified in at least 1894 knockout strains (57% of all knockout strains). Principle component analysis (PCA) was performed based on the z-score transformed abundances of the proteins across the knockout strains. Four selected self-organizing map (SOM) clusters are highlighted in different colors and contain the proteins named in the network plots c–f. b Z-scored average intensity profiles of the selected clusters outlining the changes in their expression behavior across knockout strains. c–f The selected 4 clusters are examples enriched for specific cellular pathways or subcellular localization as assigned by the STRING database (colored edges); newly identified associations are depicted with gray edges. Nodes depict individual proteins. Source data are provided as a Source Data file.