Fig. 3: Functional responses to denaturation drive heterogeneous changes in reactivity within single proteins.

a Schematic overview of peptide cluster patterns and how proteins are grouped depending on the composition of cysteine peptides from multiple clusters. The left graph shows a schematic reactivity profile of each of the four clusters. Peptides are then assigned to their parent protein, which may be deemed uni-clustered if only consisting of cysteine-containing peptides from a single cluster, or multi-clustered where cysteine-containing peptides from a single protein are associated with more than one cluster. b Mean z-score for predicted/extracted physiochemical features according to peptide amino acid composition. c Venn diagram depicting proportion of proteins for which peptides were found in each cluster combination. d Molecular weight of and e number of annotated PFAM domains in proteins to which clustered peptides are assigned. f Number of high-confidence first-shell protein-protein interactions and g inter-cluster node interaction degree annotated in STRINGdb (v11.0, score >0.7) for proteins found in each cluster. h Gene ontology terms enriched among multi-clustered proteins. Enrichment was determined using Panther GOSlim Fisher’s over-representation test with false-discovery rate correction. Common themes are denoted: A = protein folding and stress response, B = binding and complexes. Dark bars denote exact terms found to also be enriched among multi-clustered proteins in published dataset 4. Panels d–g show individual protein datapoints overlayed with mean ± S.D. Mean of combined uni-clustered proteins is shown as dotted gray line. Uni-clustered proteins were compared to those associated with multiple clusters via two-tailed t-test with Welch’s correction, *** denotes p < 0.001, **** denotes p < 0.0001, ns denotes p > 0.05. Exact n and p values are provided in Supplementary Data 1. Source data are provided as a Source Data file.