Ontological systems under development at the RIKEN BioResource Center promise to revolutionize life science research by completely changing the way researchers share data

Researchers at the RIKEN BioResource Center led by Hiroshi Masuya of the Technology and Development Unit for Knowledge Base of Mouse Phenotype are developing a system that will be able to bring together all of the information saved in databases around the world to be accessible from a single terminal. This system will allow scientists to select the information necessary for their research instantaneously from any database in the world, analyze it, and display the results in a readily usable format. The key to the system is ontology, a philosophy dating back to the time of Aristotle but with technological relevance today. Ontological systems promise to revolutionize the way we share data, and the technology is attracting attention across the globe.

Barriers to research

There are many steps researchers must take in making their own experimental plans, including checking databases to find and analyze research trends in relevant fields and choosing the appropriate experimental materials. It is also necessary to compile papers and other reference materials, and review and interpret their contents. However, regarding the issue of the wordings used to describe pathologic conditions and other characteristics of laboratory animals, for instance, different researchers use somewhat different definitions. This linguistic vagueness makes it necessary to analyze the experimental methodology and context and reinterpret the terms in all cases. A great deal of time is taken with these painstaking preparatory arrangements before determining the optimum experimental methodology.

“In biology, there are numerous databases for genes, proteins, diseases and the like around the world, and they all operate separately. A researcher who wants to investigate a particular subject must search all the databases that seem to be appropriate one by one. In addition, each individual database has its own attributes. Because the databases are designed to be used in distinct ways that are suited to different research areas, it takes a great deal of time for researchers in other areas to become familiar with databases in areas other than their own,” points out Masuya.

Ontology—correlating the essential nature of things

“Ontological technology allows computers to automatically arrange and extract the desired data so that the preparatory work for any investigation comes very easy,” Masuya explains, “The term ontology has its origin in a Greek philosophical term meaning existence. In bioinformatics, ontology refers to the classification of concepts and terms and how to describe their relationships and systems.”

In 2010, Masuya and his colleagues created the RIKEN Integrated Database of Mammals. The database incorporates YAMATO-GXO (‘Yet Another More Advanced Top- level Ontology-Genetics Ontology’), an ontology tool they developed jointly with Riichiro Mizoguchi at the Institute of Scientific and Industrial Research (ISIR) of Osaka University. “We integrated the 18 major databases of the world using YAMATO-GXO. Our mammalian database is based on RIKEN’s Scientists’ Networking System (SciNetS).” Developed by a team led by Tetsuro Toyoda, director of the RIKEN Bioinformatics And Systems Engineering Division (BASE), SciNetS can accommodate a wide variety of data, including ontological data, facilitating the integration of developed databases. To date, RIKEN’s nine databases in biology have been integrated. They succeeded in integrating as many as 900,000 data items from 18 databases by incorporating YAMATO-GXO into SciNetS and other databases. “It is quite painstaking for a single researcher to find the data they want from among 900,000 entries. However, the RIKEN Integrated Database of Mammals makes it easy to obtain the data they want in a somewhat automatically analyzed form.” This database is currently under development and expansion (see Fig. 1).

As an example, animal skin comprises a diverse range of tissues, and the presence or absence of hair is a good indicator for detecting tissue anomalies. ‘Nude’, ‘hairless’, ‘hair loss’ and ‘loss of hair’ are all terms used to describe the absence of hair on the skin. The most appropriate term to use depends on the audience and the journal’s editorial policy. Using a conventional database, for example, the term ‘nude’ may get 18 hits concerning related genes, whereas ‘loss of hair’ might get only two hits. “All these terms share the essential meaning of ‘no hair on the skin’ and ontology helps to determine the essential meaning of words and sentences and to correlate a broad range of things and words. It arranges and systematizes biological concepts and the connections with their essential meanings. When we search in the Integrated Database of Mammals with a retrieval term such as ‘nude’ or ‘loss of hair’, all the relevant genes are retrieved and displayed instantaneously.” Because the logical structures, including concepts, information and data intrinsic to those individually existing databases, have been integrated into YAMATO-GXO, a unified knowledge structure for biology, with accurate search and extraction functions has become possible.

Rapid integration of databases

Technology for knowing the meanings of words and sentences and forming correlations among them may sound relatively simple, but ontology is in reality a very profound activity. “Ontology is philosophy. It is underlain by a philosophical system that has been unbroken since the time of the ancient Greek philosopher Aristotle (BC 384–322). It took five years for us to be able to understand information technology based on a philosophy that has been nurtured over such a long historical period.”

According to Masuya, “Ontology is used to teach the computer about this world.” For example, the human being is a primate, a mammal, an organism and an animal. It is characterized by bipedal locomotion, a brain weighing 1,250 grams on average, five fingers on each limb, two eyes and so on. By fractionalizing things like this and systematizing the essential meanings, a more fundamental ‘superordinate concept’ is created (Fig. 2). If systematized, even databases with different logical structures can be combined relatively easily with ontology serving as a ‘translator’.

Before ontology was integrated into practical applications, databases could not be linked together unless their logical structures, or intrinsic habits, were coordinated in all cases. The need to build other databases to separately connect different meanings was unavoidable. That work is painstaking and time-consuming. “Thanks to YAMATO-GXO, we were able to develop the RIKEN Integrated Database of Mammals, which integrates 18 databases, in just half a year.”

Building on this achievement, in fiscal 2011 RIKEN launched the ‘Biological and Environmental Phenomes Integration Database’, a database integration promotion program sponsored by the Japan Science and Technology Agency in a joint initiative with Toyoda of the BASE. “This program will integrate nationally available data on ‘phenotypes’, which represent the characters manifested by the action of genes, and information on measurement techniques. We are working on developing a database that allows even a measurement technique with use limited to a particular area to be used in other areas, allowing it to contribute to advances in biology at large.”

The attraction of ontology

Ontology research is currently attracting worldwide attention. The concept of gene ontology was first proposed in 1995 by Michael Ashburner of the University of Cambridge in the UK, and gene ontology even now represents a major technical breakthrough for the standardization and massive compilation of biological information. The introduction of this approach resulted in an explosion in research using DNA microarrays—chips that allows investigators to determine how a large number of genes are expressed, and the intensity of expression, at one time. Using gene ontology, for example, it is possible to collate the availability of all reports on the functions of the gene expressed. With the spread of DNA microarray technology, a new discipline called transcriptomics emerged to analyze when, where and at what levels the more than 20,000 human genes are expressed, and to determine what is meant by the expression. “The microarray could not have become such a powerful research tool without gene ontology,” says Masuya. Linkage of the two distinct technologies, microarrays and ontology, has been promoting advances in the new research domain of transcriptomics.

The trend of the times is also boosting ontology. It has been shown that in research into genes and proteins, causality does not always stand in a one-to- one relationship between cause and result. This is because many genes and biomolecules are involved in the processes for the generation of each protein. Additionally, techniques for visualizing the behaviors of many genes and biomolecules are already available. “By using an ontology-based integrated database, we can get a listing of the results from the concurrent functioning of multiple genes out of the vast amount of data obtained, rather than the one-to-one matched data on gene functions in conventional databases. Ontology is expected to really lead future research.” Because it is capable of easily identifying disease-causing genes and proteins from among the vast number of biomolecules, ontology is expected to lead to major breakthroughs in the acceleration of new drug development and phase I clinical trials. While information is increasing explosively in the research domain, ontology that links a wide variety of databases can be described as a hidden but powerful tool that leads research activities that are prone to become chaotic.

Identifying knockout mouse phenotypes within an international framework

“We will proceed to develop ontology to standardize international mouse information,” says Masuya. His laboratory has been requested to join the International Mouse Phenotyping Consortium (IMPC) to clarify the relationships between genes and phenotypes by examining all the phenotypes in knockout mice that have been manipulated to systematically delete each gene in the mouse genome. Mice represent a number of similarities (homologies) with humans in terms of the number and kinds of genes, as well as biological events and disease processes. The large project aims to link human diseases and phenotypes of knockout mice. “Currently, laboratories all over the world are working to design knockout mice and utilize them as investigational materials independently. However, a major loss of information resides here.”

In conducting experiments, researchers create knockout mice that fit their research themes. For example, researchers studying limb development may generate a knockout mouse by inactivating a relevant gene. If researchers cannot find any morphological abnormality in the limb, they often give up on investigating that mouse further. However, a lot of genes have multiple functions. For instance, many signaling molecules involved in limb development are also involved in other biological processes in another organ—a fact that could be easily overlooked and a discovery that might never get published, even though the finding may have made all the difference to a physician struggling to elucidate metabolic disorders in a patient.

“Such occurrences have been prevalent since the birth of the first knockout mouse. The IMPC offers a decisive solution to this situation.” In the large- scale project with its huge budget of nine million dollars, more than 20,000 mouse genes are being knocked out one-by-one to comprehensively analyze basic phenotypes and determine their influences on the mammalian body. The project also includes the development of an ontology-incorporating database and provides free access to information on the associations of the genes with biological phenomena and diseases. RIKEN’s BRC is going to join the IMPC in a collaboration between the Technology and Development Team for Mouse Phenotype Analysis led by Shigeharu Wakana, the Experimental Animal Division headed by Atsushi Yoshiki, and Masuya’s Technology and Development Unit for Knowledge Base of Mouse Phenotype. Once this information network is built, it will be possible to list all knockout mice that help research into a particular human disease from the database. “The network will enable us to select ‘all’ mice serving as disease models that exhibit similar symptoms, and even ‘potential models’ that exhibit near-morbid conditions. This encompassing ‘all’ is of paramount importance, and reducing the unidentified portions will dramatically move forward the whole field of research into disease.”

A powerful tool that will lead research activities

Database integration using ontology has the potential to bring many breakthroughs. In a hospital, for example, physicians could download a listing of everything from the names of candidate diseases to the likely progression of the condition, candidate medications and therapeutic guides. Such an integrated database would make it possible to investigate therapeutic approaches to coping with complications from all angles using information from the component databases. “Our ultimate goal is to create a tool that will serve as the guide to researchers’ activities by presenting information even at levels beyond human ponderings, and deducing and displaying potentially useful search results in an easily understandable way,” says Masuya.

Hiroshi Masuya
Unit Leader
Technology and Development Unit for Knowledge Base of Mouse Phenotype
RIKEN BioResource Center {hwdvs-related}number=3|keywords=network,database,biology{/hwdvs-related}