Project Summary/Abstract Reducing advanced breast cancer mortality requires urgent development of better drugs and improved therapeutic strategies; however, new drug development is extremely time-consuming and costly. With the explosive growth of large-scale cancer genomic and phenotypic data (e.g., the Cancer Genome Atlas [TCGA]) and publicly available high-throughput screening data for thousands of small molecules (many of which have already received regulatory approval for at least one medical condition), computational drug repositioning or repurposing holds great potential for precision medicine and may provide tools to significantly improve breast cancer treatment and outcomes. Our hypothesis is that optimal therapeutic choices can be identified for hard to treat breast cancers by applying transcriptome-based drug sensitivity prediction methods. Our long term goal is to identify and validate the efficacy of existing drugs in hard to treat breast cancers, namely triple negative breast cancer (TNBC) and metastatic breast cancer (MBC). Toward this goal, this proposal contains two specific aims to develop, apply, and improve methods to predict drug sensitivity (either as a single agent or in combination). We will also validate these predictions in additional large-scale cancer genomic datasets and translate the results using cell based and in vivo (mouse) models of TNBC and MBC. In Aim 1, we will focus on identifying effective drugs as monotherapy, while Aim 2 is to identify and validate optimal therapeutic combinations. Our study is significant because it will accelerate the development of novel therapies for hard to treat breast cancers by repurposing existing drugs, thus avoiding the lengthy and risky new drug development process. The ability to tailor therapy for specific disease subtypes and identification and validation of new drug indications will provide valuable therapeutic options in the battle against TNBC and MBC, and subsequently reduce their associated mortality. Our proposed research is innovative in both the methodologies employed and their applications, as our transcriptome-based drug sensitivity prediction represents a paradigms shift in drug sensitivity prediction; furthermore, we are applying these novel prediction approaches to patient tumor data not only for biomarker discovery in order to tailor individual therapy, but also for drug repurposing. The ability to bring biomarker discovery and drug repurposing together will present a new opportunity for cancer therapy, as the whole genome expression profile of a tumor will be used to provide optimal therapeutic options in different cancers, and many “old” drugs can find a new purpose in improving cancer treatment outcomes.