Fig. 3: The cell cycle phosphoproteome is characterised by intrinsic disorder and MLO components.

a Scheme illustrating hypothetical enrichment of phosphorylation in disordered regions when taking into account amino acid compositional bias. b Scatter plot of expected vs observed phosphorylated Ser/Thr for each protein of human and Xenopus phosphoprotein datasets. FDR thresholds of 5% and 1% are marked in yellow and red respectively. Circles: proteins with at least one dynamic phosphorylation in Xenopus, or human CDK1 subfamily substrates, respectively. c Boxplots showing expected vs observed phosphorylated Ser/Thr in IDRs (IUpred) among all phosphoproteins detected (left), phosphoproteins with at least one dynamic phosphosite (middle), and dynamic phosphoproteins also detected as CDK1 subfamily targets in humans (right). Distributions were compared with the paired Wilcoxon signed-rank two-sided test. All phosphopoteins (1843), p < 2.2e−16; Dynamic (646), p < 2.2e−16; Dynamic and human CDK target (147), p = 1.0e−14. Boxplots centre, median; lower and upper edges, 25% and 75% quartiles, respectively. Whiskers, data with the largest or smallest values not further than 1.5* interquartile range (IQR) from the upper or lower box limits, respectively. Beyond these values, data were not plotted, for clarity, but were included in the statistical analysis. d Plots showing the common Odds Ratio of Ser/Thr phosphorylation in structured and disordered regions calculated with the Fisher’s test (see Supplementary Fig. 5b, c for the statistical analysis scheme and exact p values). For all organisms, the disordered regions were calculated with three different disorder predictors. The disordered fraction is presented in a colour scale. e Violin plots of the distribution of disordered residues per protein for CDK targets vs the rest of the phosphoproteome for human and yeast, and dynamic phosphoproteins vs the rest of the phosphoproteome for Xenopus. Intrinsic disorder was calculated with three different predictors (IUPred, SPOT, and VSL2b). Statistical significance was evaluated with the Wilcoxon–Mann–Whitney signed-rank two-sided test: Xenopus (CDK targets, 646, Non-CDK targets, 1198): IUpred p = 2.36e−13, SPOT p = 4.83e−13, VSL2b p = 5.78e−10; Human (CDK targets, 16167, Non-CDK targets, 2153): IUpred p < 2.2e−16, SPOT p < 2.2e−16, VSL2b p < 2.2e−16; Yeast (CDK targets, 100, Non-CDK targets, 2153): IUpred p < 2.2e−16, SPOT p < 2.2e−16, VSL2b p < 2.2e−16. Boxplot parameters as in (c). f Violin plot (left) showing the distribution of disordered residues per protein (as estimated with SPOT) for CDK, MAPK, Aurora, PLK, NEK and DYRK kinase targets vs the rest of the phosphoproteome for human targets. Statistical significance was assessed by Kruskal–Wallis ANOVA, and pairwise comparisons were performed with Dunn’s two-sided post hoc tests. The adjusted p-values (Benjamini–Hochberg) are shown in a tile plot (right). Boxplot parameters as in c. Source data are provided as a Source data file.