We introduce a database platform called `RIKEN Meta Database' that provides information on RIKEN's various life-science databases to help researchers around the world make full use of RIKEN's research results. Since RIKEN procures various and large life-sciences datasets through the genomes and phenomes of various species, including datasets for sequence and image data, such resultant research data and databases are heterogeneous and tend to be published in data-domain speci c interfaces and formats. To achieve data re-usability, including facilitating data/database discovery and data description as life-science knowledge for further intelligent data analysis, the database platform is implemented to manage metadata of both data and database using Semantic Web technologies with standardised vocabularies and life-science ontologies. As of September 2015, 153 databases, including external public databases and 15 ontologies, are integrated on the platform, and these data can be access via a SPARQL endpoint and a browser-based graphical user interface that shows metadata in table and graph format.
In recent life-science research, theses research elds have become deeper and more specialised, and collaboration among di erent research elds, such as informatics and material sciences, are actively promoted to comprehensively understand life phenomena. RIKEN is a comprehensive science institute with research centres and laboratories, including physics, chemistry, informatics and life sciences laboratories, that produces various and large data as individual databases. Even in RIKEN, which represents the science community in microcosm, such data and databases are di cult for researchers to nd, and intelligent and integrated analyses has not been realised. To address this problem, we have been developing a database platform called the `RIKEN Meta Database' (http://metadb.riken.jp) to facilitate the management and publication of database metadata according to the Semantic Web. The principal databases published by RIKEN have generated metadata that can be accessed by SPARQL and a browser-based graphical user interface (GUI). 2
The RIKEN Meta Database is designed to be a lightweight database platform. The core components of the platform are 1) a SPARQL endpoint, 2) a GUI web server that provides a GUI for data input and display and 3) a resource description framework (RDF) data converter. The SPARQL endpoint manages RDF data and is a server for a user's SPARQL client and the GUI web server. The GUI web server provides reader-friendly web pages for each database, class and entities in table or graph format by accessing the SPARQL endpoint. The web server also provides an interface that accepts RDF datasets for data uploads. The data converter is a Java application that accepts a Microsoft Excel le, which is converted to RDF format. This converter is used for wet researchers who are willing to publish their data but are not familiar with the RDF format. Currently, components 1) and 2) have been deployed on Linux virtual machines with a 1 GB ash memory drive having 96 GB and 8 GB memory, respectively. 3
As described, RIKEN is a comprehensive science institute that can perform
data editing in close cooperation with biologists and informaticians. In such an
environment, careful selection of public ontologies and data classes has been
performed for each database. As of September 2015, with 15 public
ontologies and 153 databases, including RIKEN's principal databases such as
`FANTOM' (mammalian [