Infectious diseases remain a formidable threat to global public health, with their escalating morbidity and mortality rates compounded by recurrent epidemics and the alarming rise of antimicrobial resistance (AMR). These challenges have intensified the urgent demand for innovative therapeutic strategies that can accelerate drug development cycles and overcome traditional research bottlenecks. To address these critical needs, we present IDKG (Infectious Disease Knowledge Graph), a specialized large-scale biomedical knowledge network designed to bridge data fragmentation through multimodal data integration. The IDKG constructs comprehensive associations from 345 infectious diseases and 708 pathogens across heterogeneous biomedical sources systematically. The graph architecture comprises nearly 50,000 nodes (8 types, including Pathogen, Protein, etc.) and over 1.2 million edges (11 types, including treats, contains, etc.), establishing an interconnected framework that enables systematic interrogation of cross-disciplinary knowledge. The integrative approach effectively dismantles conventional data silos while preserving biological contextuality. We validated the IDKG\'s potential by applying graph neural network-based approaches for drug repurposing prediction in human metapneumovirus (hMPV) infection, a common acute respiratory infection for which effective specific antiviral drugs are currently absent. The successfully identification of established antiviral agents, such as ribavirin and emetine, by our M1 model demonstrated its predictive accuracy and biological relevance. IDKG unifies multimodal biomedical data into a network to accelerate drug discovery and bolster outbreak response. This establishes a data-driven, knowledge-based paradigm for infectious disease research.