Personalized oncology requires accurate access to complex evidence from gene-drug interaction. We developed the GeneDrug Platform to address this challenge, through a dual-source hybrid architecture that provides physicians with two complementary pathways: one to find high-frequency mutation-drug regimens from authoritative clinical guidelines and another to discover low-frequency mutation treatments from PubMed literature. This system integrates a high-fidelity Knowledge Graph constructed from more than 1,000 cancer guidelines with a real-time LLM-driven relation extraction engine accessing PubMed literature. It can substantially reduce the time for physicians to formulate treatment regimens and provide fully traceable evidence sources to support clinical verification. The proposed system is publicly available at https://genedrug.org .