Heterogeneity of genomic instabilities among individual patients is believed to be a major cause of drug response heterogeneity. Cancer patients who are sensitive to anti-cancer drugs are often re-examined to understand the unknown mechanism of action (MoA) of given drugs. For example, a non-small cell lung cancer (NSCLC) patient was reported to be responsive to Dasatinib treatment and remained cancer-free four years later. Though follow-up genomic analysis showed the patient bears an inactivating BRAF mutation in the tumor, the MoA remains unclear. There are two challenges in uncovering the MoA. First, Dasatinib is a kinase inhibitor, which often has many protein targets. Second, the downstream MoA signaling pathways regulated by these targets are too complicated to delineate. Currently, there is no computational tool that can effectively address both challenges. To fill this gap, we developed a computational tool DrugMoaMiner (Drug MoA Miner) that can be used to identify the comprehensive set of kinase inhibitor targets; delineate the underlying drug MoA; and predict personalized sensitivity to a given drug based on an individual's gene expression profiles. We applied the DrugMoaMiner to lung cancer cell lines to uncover the potential MoA signaling network of Dasatinib sensitivity; our result is in agreement with previous protein data analysis. Moreover, we can predict Dasatinib response of an independent set of NSCLC cell lines using the MoA signaling network uncovered.