) using strict control for false identifications. Each site has been classified into the Gold-Silver-Bronze confidence categories. First, we identified 55 significant motifs and grouped these into kinase classes to perform pathway enrichment analysis. Next, we leveraged disorder region predictions and AlphaFold 3's ability to consider post-translational modifications (PTMs) when modeling proteins to understand the structural context of phosphosites. Here, we determined that phosphorylation tends to occur on disordered serine and threonine residues. AlphaFold predictions suggest that phosphosites induce alpha helices to form in proteins, although many "induced helices" appear to be unusually short and require further validation. As artificial intelligence (AI) is being applied in proteomics, we must ensure that publicly available data are accurate and of high-quality to be used for downstream analyses and training models. With this motivation, our results are available in PRIDE (PXD071918), PeptideAtlas and UniProtKB, ensuring that this PTM data is FAIR and "AI-ready".
Boswell et al. (Mon,) studied this question.