Inferring Kinase Activity from Tumor Phosphoproteomic Data

University Of Virginia, Charlottesville VA

Investigators

Abstract

Project Summary Kinases are fundamentally important enzymes for regulating cell physiology through regulation of proteins and protein interactions by phosphorylating tyrosine, serine, and threonine residues. Kinase dysregulation is often a contributor to cancer progression, which is why kinase inhibitors are one of the largest classes of FDA-approved drugs for oncology. However, many challenges still remain in providing precision-based kinase therapy to pa- tients, such as failure to respond to therapy and the development of resistance to therapy through diverse means. This project seeks to advance a promising new approach (called KSTAR) for understanding kinase dysregulation in cancer by inferring the activity of kinases in tumor biopsies, based on their phosphoproteomic profiles. KSTAR is a first-in class algorithm that can operate on any type of phosphoproteomic data, not requiring paired quantita- tive comparison tissues, and is significantly more robust than other available approaches. KSTAR was shown to compliment clinical standard of care by identifying failure to respond to therapy and misclassification of patients as HER2-positive or negative, which departed from HER2-activity. Working with collaborators across a range of solid cancers, we seek to further KSTAR's ability to help researchers and clinicians better match kinase inhibitor therapies, based on patient molecular kinase activity profiles. Key algorithmic improvements will be performed, such as: expansion of the approach to cover all human kinases, deconvolution of signaling from immune and stroma components of a solid tumor biopsies, and increasing speed. This work will advance and harden dissem- ination of KSTAR across a variety of platforms that will allow maximum flexibility for other programmers, but also web-based interfaces that require no programming to interact with patient and cell kinase profiles.