This paper proposes an immune-based framework for adaptive query expansion in the semantic web, where exploratory queries, common in biological information retrieval, can be answered more effectively. The proposed technique has a metaphor with negative selection and clonal expansion in immune systems. This work is differentiated from the previous query expansion techniques by its data-driven adaptation. It utilizes target databases as well as the ontology to expand queries. This data-driven adaptive feature is especially important in exploratory information retrieval where the querying intention itself can be dynamically changed by relevance feedback.