Project Summary Despite the progress made in precision oncology, clinicians typically face a vast volume and variety of next- generation sequencing and molecular data that is frequently intuitively processed to support high-stakes decisions. Overall, currently available resources that assist with next-generation sequence data interpretation are limited by manually performed, complex, time-consuming, and error-prone gene queries and ultimately lack the necessary information for prioritizing emerging therapies in a scalable manner. Importantly, the integration of genomic with clinical data has been severely hampered by the lack of advanced analytical tools that match genomic targets with molecularly-driven therapies. These barriers, together with health disparities, widen the gap between an exponentially increasing drug development field and the actual benefits for patients with cancer. The overarching goal of the proposed research is to link clinical with computational precision oncology and enable clinical decision-making in genomically defined groups. We propose to develop a precision oncology decision support framework for automated, scalable, and precise matching of actionable next-generation sequencing findings with targeted therapies. We will then test its clinical utility and value in the several clinical settings within the Johns Hopkins Molecular Tumor Board, in Johns Hopkins partnering community medical centers as well within two ongoing clinical trials for women with breast cancer. To enhance the generalizability of our analytical toolkit past our local academic environment, we have designed the platform's architecture such that it allows for ingestion and harmonization of next-generation sequence data from multiple sources, implements a common data model to map clinical elements to standardized terminologies and leverages ensemble natural language processing to generate actionable mutation-targeted therapy pairs. These attributes provide the foundation for the toolkit's potential widespread use and implementation in health care settings outside our local academic environment. While significant advances have been made in advanced diagnostics for tumor profiling, a solid backbone that supports the practical implementation within and across health care systems is lacking. The underlying premise of the proposed research is that it will ignite cross-institutional real-world genomic data analysis initiatives and genotype-driven clinical trials that will be beneficial for health systems and patients. Notably, our precision oncology decision support platform will enhance the implementation of precision oncology at institutions that do not readily have access to in-house expertise in clinical genomics. We envision that this streamlined automatic and scalable process will improve care, enhance patient outcomes and define national standards in how treatments are selected and tailored to individual patients.

Project Narrative We propose to develop a fully automated precision oncology decision support system that unifies mutation characterization by levels of evidence, provides links to targeted therapies and via a user frontend can be integrated in clinical decision making in academic institutions and community medical centers.