Clustering of phosphorylation changes
Phosphorylation sites with one-to-one mapping to human S, T or Y amino acids and showing significant change in phosphorylation (abs(log2FC) > 1 and adjusted p value < 0.05) in one or more conditions were selected and clustered using hierarchical clustering (complete-linkage) with pearson correlation (1-r) as the distance measure. The cluster tree was cut into 5 clusters using a dynamic tree cutting method, cutreeHybrid function in dynamicTreeCut (Langfelder et al., 2008) package in R, with a minimum cluster size of 130 sites. Phosphorylated proteins in each cluster were tested for enrichment of Reactome pathways (Jassal et al., 2020). The over-representation analysis (ORA) was based on the hypergeometric distribution and performed using the enricher function of clusterProfiler package in R (Yu et al., 2012). The pathway terms were obtained from the c2 category (Reactome) of Molecular Signature Database (MSigDBv6.1).