The advances of microarray technology have made large amount of gene expression data available from a variety of different experimental conditions. Analyzing the microarray data plays a key role in understanding gene functions, gene regulation and cellular process. Order-Preserving Submatrix (OPSM) is an important model in microarray data analysis, which captures the identical tendency of gene expressions across a subset of conditions. In the process of analyzing mechanism of gene expression, OPSM search undoubtedly saves the time and effort of biologists. However, OPSM retrieval mainly depends on keyword search, resulting a weak control on the obtained clusters. Typically, the analyst can determine the ad-hoc parameters which are far from the declarative specification of desired properties on operation and concept. Motivated by obtaining much more accurate query relevancy, this paper proposes two types of OPSM indexing and constrained query methods based on signature and Trie. Extensive experiments conducted on real datasets demonstrate the proposed methods have better behaviors than the state-of-the-art methods on efficiency and effectiveness.