Because of advances in biotechnologies, various types of biological data have drastically increased in the past decade. Because of the volume, heterogeneity, and continual growth of biological data, it has become increasingly difficult for individual researchers to manage an entire dataset in a single repository. In this context, Semantic Web technology [1] has attracted attention as a promising approach of knowledge management [2]. In the Semantic Web, all information is described in the Resource Description Framework (RDF) [3], in which every piece of information is in the form of a triple containing a subject, predicate, and object and each resource is represented by a Uniform Resource Identifier (URI). The RDF works as a general framework of knowledge representation and the URI assures valid integration of data collected from different sources. Furthermore, knowledge extraction from the RDF can be implemented using a powerful query language called the SPARQL Protocol and RDF Query Language (SPARQL) [4]. SPARQL specifications include federated query functionality [5], by which distributed databases can be queried in an integrative manner. Thus, Semantic Web technology provides a basis for flexible integration of the increasing amount of heterogeneous data. In fact, many biological databases have already adopted the Semantic Web [6–9].

Despite the well-designed basis of Semantic Web technologies, several obstacles that prevent users including bioinformaticians from utilizing RDF databases still remain. The main hurdle for most users is writing SPARQL, which often includes cumbersome coding tasks. For example, SPARQL permits inclusion of subqueries for distinct endpoints in a federated query; however, writing such a nested query is a complicated task and can be a technical obstacle for most users. Several approaches for supporting SPARQL coding currently exist. Examples include SPARQL editors with useful functionalities such as URI autocompletion [10], and graphical support for step-by-step construction of SPARQL queries [11, 12]. Despite these approaches, constructing executable SPARQL code, even for a simple query, still remains a time-consuming task; thus, a mechanism that saves time of preparing SPARQL code is necessary to maximize the use of available RDF datasets. As an alternative approach to this issue, a wiki-based portal for sharing SPARQL queries was constructed [13], which can bypass the burdensome coding task. Although the queries registered on this service can be executed on the portal site, a mechanism for reusing these queries in other environments would maximize the usefulness of the accumulated queries.

Here, we developed SPANG, a client that supports querying by generation and reuse of SPARQL codes through a simple interface. Taking advantage of the common “triple” form of RDF data, SPANG generates typical queries without the need for SPARQL coding. Even in complicated queries, SPANG can construct runtime queries using predefined templates. Regarding the federated query, SPANG realizes a similar functionality by combining multiple queries through a Unix pipe. SPANG, with its unique features, minimizes the burden of coding SPARQL, thereby enhancing integrative exploitation of distributed databases.

Although each spang process submits a query to a specified database, the spang process can be combined with other Unix processes through a Unix pipe. Notably, multiple spang commands, each with distinct target database, can be combined through a Unix pipe by transferring variable bindings between queries, thereby realizing federated use of multiple databases.

The SPANG package is implemented in Perl. Specifically, the spang command accesses remote SPARQL endpoints using the Perl LWP module. To lower the initial hurdle of querying with SPARQL, the SPANG package provides predefined configurations, including i) nicknames for SPARQL endpoints, ii) frequently used prefix declarations for URIs, and iii) SPARQL template libraries. Furthermore, users can extend the configurations by preparing user-defined configuration files.

In this paper, we presented SPANG, a SPARQL querying client that has several unique features. First, SPANG provides a shortcut mode that can generate a simple query containing a certain triple pattern. This mode aids querying with SPARQL and is helpful for beginners to start exploring RDF datasets. It is also useful for experienced users of SPARQL, as useful information can often be obtained by retrieving adjacent nodes in RDF graphs using the shortcut mode and efficiently submitting such simple queries is crucial in data mining. Second, for more complicated queries, SPANG provides a template mode, by which existing SPARQL codes can be reused among users. This mode enhances the usage of SPARQL through development of SPARQL template libraries that represent reusable query patterns. The template libraries constructed by experienced users can help other users to efficiently utilize RDF databases. Third, the queries in either shortcut or template mode can be combined in the Unix command line to realize a more complex query. This modular structure of queries has several merits: it reduces complexity of each SPARQL query, leading to easier implementation and debugging of the query; and it extends potential application of each query through combination with other queries or Unix commands.

The predefined SPARQL templates included in the SPANG package are available to help users query some biological RDF databases. However, the range of queries included in the package is limited to rather common ones. The potential use of SPANG can be further extended by database users or database providers through development of SPARQL template libraries. Although a service for sharing SPARQL queries exists [13], it is difficult to execute them directly for instant reuse by users. In SPANG, users can directly call SPARQL templates across the Web. Thus, if an RDF database provider, who knows best the manner in which the database should be used, publishes SPARQL template libraries, database usage can be considerably enhanced. This study suggests the possibility of an open framework of sharing query in a reusable form. Future work may include the standardized use of the query templates, which will further facilitate the sharing of useful queries. Sharing not only data but also queries (i.e., means of interpreting data) on the Semantic Web platform will help the biological research community collaborate in knowledge integration and discovery.

SPANG enables easy generation of typical queries, thereby reducing the burden of writing SPARQL. SPANG also provides a framework for reusing and sharing arbitrary queries across the Web. Moreover, it enables users to execute complex queries by combining existing query templates. SPANG, with these unique features, facilitates integrative exploitation of published RDF datasets and supports knowledge discovery.