Rare disease drug development faces significant challenges due to small patient populations, fragmented evidence, and limited commercial incentives, making drug repurposing an attractive strategy when genetic, pathway, pharmacological, and regulatory evidence are considered together. Existing computational repurposing approaches often focus on biological similarity or drug–target relationships while treating regulatory knowledge as external context, which restricts the ability to prioritize candidates with orphan drug precedent, approved indications, or interpretable translational relevance. To address this, we propose a knowledge-graph framework that integrates gene–disease associations, biological pathway relationships, drug–target evidence, and orphan drug regulatory labels, enabling the generation and ranking of repurposing hypotheses for rare diseases without claiming experimental validation or quantitative performance. The system incorporates entity-resolution workflows, a graph database, a reasoning layer, and a hypothesis-ranking module, linking disease, gene, pathway, drug, orphan designation, and label-indication entities while preserving provenance for each relationship. This framework allows researchers to query for drugs that target pathways disrupted by rare disease genes and that have orphan designation or approval evidence in related conditions, making mechanistic and regulatory evidence visible within the same reasoning environment. By embedding orphan drug regulatory knowledge into a biological knowledge graph, the approach helps bridge the translational gap between mechanistic discovery and clinical development, supporting transparent, evidence-aware prioritization of repurposing opportunities for rare diseases.