Drug repositioning, the discovery of new indications for existing drugs or biologics, provides an alternative way to designing new drugs. Despite not a new concept, it is not an easy task to identify possible new indications for old drugs. In this paper, we propose a new framework to identify drug candidates against breast cancer, where these agents are not originally developed for breast cancer. In particular, we assume that the drugs used to treat breast cancer should reverse the activities of signaling pathways affected by the disease. Specifically, we present an optimization model to identify signaling pathways affected by causal genetic mutations of breast cancer from molecular interaction networks. Applying our method to real datasets of both breast cancer and bioactive compound perturbations, we successfully identify some drugs that can be repositioned for breast cancer as reported in literature, which demonstrates the effectiveness of our proposed approach.