We develop a computational-experimental strategy, based on information theory that resolves the intra-tumor molecular heterogeneity on the single cell level and allows to rationally design patient-specific treatments. The approach breaks down tumors into the subpopulations that they possess, and altered protein networks associated with each subpopulation. The resolved patient-specific intra-tumor heterogeneity is used to predict and test the optimal treatment. Moreover, these networks can be used to rationally plan personalized sensitization of TNBC tumors to RT/ CT.