Conventional mass spectrometric proteomics methods don’t work well for low-abundance proteins in complex mixtures. Edward M. Marcotte, Eric V. Anslyn, and coworkers at the University of Texas, Austin, report a method that allows them to perform highly parallel single-molecule identification of proteins at zeptomolar concentrations (Nat. Biotechnol. 2018, DOI: 10.1038/nbt.4278). To do this, they harness a classic protein-sequencing method—Edman degradation. In Edman sequencing, amino acids are chemically clipped from peptides one at a time and identified via chromatographic separation. In the new method, the researchers skip the chromatography and use fluorescence microscopy to look at the peptide left behind. They immobilize peptides that have been fluorescently labeled at lysine and cysteine residues and use a microscope to detect the fluorescence from the labels. They then subject the peptides to multiple rounds of Edman degradation. When a fluorescently labeled amino acid is clipped off, the peptide’s fluorescence either decreases or disappears entirely. Because the researchers know how many cycles occurred, and thus how many amino acids were clipped, they can determine where the labeled amino acid was located on the peptide. Although they demonstrated labeling of only cysteine and lysine, the researchers also have reagents that target glutamate, aspartate, and tryptophan. Computer simulations suggest that those five amino acids plus the specificity of protein-cutting enzymes should provide enough information to identify most proteins in the human proteome.