CysNet: Theorem constrained inference of cysteine redox proteoform states from bottom-up mass spectrometry data

Here, we present CysNet, a theorem-constrained method designed to infer cysteine redox proteoforms (oxiforms) from bottom-up, mass spectrometry (MS) based proteomic data. This overcomes limitations with previous MS redox proteomic approaches, which can quantify residue-resolved cysteine redox states, but leave distinct oxiforms unresolved. CysNet treats each residue-resolved oxidation value as a binary redox-coordinate marginal, enabling theorem-constrained inference of the oxiforms that are necessary, impossible or bounded within the compatible protein-group ensemble. This collapses the vast theoretically possible set of oxiform states to a finite set of allowed values by extracting existence and exclusion constraints from the data, despite the incomplete proteome coverage typical for bottom-up MS datasets. Using CysNet to analyse human induced pluripotent stem cell lines (6,300 cysteine-containing protein groups, 22% cysteine coverage), resolved 519 exact oxiforms, inferring 7,000 oxiforms per line. Quantitatively, CysNet bounded the oxiform content to approx. 15% of the measured cysteine proteome. These data define the deepest oxiform survey recorded. CysNet revealed a latent structural layer in redox variation between the cell lines, distinguishing changes in oxiform identity (composition) from changes in oxiform weighting (intensity). Hence, CysNet moves bottom-up redox proteomics beyond isolated site-level cataloguing by reconstructing copy-number-weighted oxiform maps, providing a scalable route to deep oxiform information from peptide-level data.