Data set II: Proteome and phosphoproteome in CPTAC ovarian cancer cohort

In the second case study, we visualize high-grade serous ovarian carcinoma data from mass spectrometry-based untargeted proteomics and phosphoproteomics experiments conducted by the Clinical Proteomics Tumor Analysis Consortium (CPTAC). Our visualization follows that of Zhang et al²². In their analysis, the authors retained the 3586 proteins out of the 9600 that were quantified in 169 tumors. For the phosphoproteome data, the authors quantified the relative abundance for 69 tumor samples. Among these, 67 samples were quantified at both omics level and were used in the final visualization. Phosphosites with more than 50% of missing data were filtered out and the remaining missing values were imputed using KNNImpute⁶⁴. Since ischemia of the TCGA tumor samples was found to be a confounding variable that altered phosphopeptide abundance, phosphosites that were shown to be regulated in ovarian carcinoma⁶⁵ were also removed. The abundances were converted to z-scores before visualizing in multiSLIDE. A total of 16718 kinase-substrate interactions were curated from PhosphoSitePlus³¹, PhosphoNetworks³² and a predictive network inference approach³³ to build the kinase-substrate map.