Computational analysis of cholangiocarcinoma phosphoproteomes identifies patient-specific drug targets

We used mass-spectrometry based phosphoproteomics and computational methods to identify patient-specific drug targets in primary CCA and CCA-derived cell lines. We analyzed 13 primary CCA with matched background tissue and 8 different cell lines leading to the identification and quantification of >13,000 phosphorylation sites. Application of the Drug Ranking Using Machine Learning (DRUML) algorithm identified inhibitors of HDAC and PI3K pathway members as highly ranking in primary CCA relative to background. The accuracy of drug rankings based on predicted responses was confirmed using cell line models of CCA. Together, our study uncovers frequently overactive biochemical pathways in primary CCA and provides a proof-of-concept of the use of machine learning for ranking drugs based on efficacy within individual patient tumors.