Molecular profiling ? the practice of molecularly testing tumors in order to find specific gene or protein alterations which can be used to recommend targeted therapies ? is increasingly used in oncology and is fast becoming a major part of precision medicine. In practice, each patient or physician generally receives a list of molecular anomalies and a list of therapies which are predicted to be beneficial or not beneficial based on the tumor molecular profile. This may lead to a difficult process of prioritizing therapies for individual patients. In the current proposal, we will develop computational network-based approaches to therapy recommendation by using existing resources to inform the connections between drugs and gene or protein variants. We will consider approaches both for creating ?average? networks based on population-level data and for creating ?patient-specific? networks based on an individual?s specific tumor profile. We will also design and implement an interactive data visualization approach for these networks which will be usable by both clinical researchers and clinicians. The methods and tools developed as part of this project will be entirely reproducible and shared with the community via open-source software packages and interactive tools. We believe our project could eventually lead the way to improving the way therapies are targeted to cancer patients.
Molecular testing of cancer patients? tumors is a key component of the movement towards personalized cancer therapies by allowing physicians to prescribe therapies targeting specific gene or protein alterations in a patient. However, this approach may result in a long and potentially contradictory list of treatments which are difficult to prioritize. In this proposal we will develop methods for building networks connecting drugs to gene and protein variants in order to improve therapy recommendations for cancer patients.
