Establishment and application of information resource of mutant mice in RIKEN BioResource Research Center

Abstract

Online databases are crucial infrastructures to facilitate the wide effective and efficient use of mouse mutant resources in life sciences. The number and types of mouse resources have been rapidly growing due to the development of genetic modification technology with associated information of genomic sequence and phenotypes. Therefore, data integration technologies to improve the findability, accessibility, interoperability, and reusability of mouse strain data becomes essential for mouse strain repositories. In 2020, the RIKEN BioResource Research Center released an integrated database of bioresources including, experimental mouse strains, Arabidopsis thaliana as a laboratory plant, cell lines, microorganisms, and genetic materials using Resource Description Framework-related technologies. The integrated database shows multiple advanced features for the dissemination of bioresource information. The current version of our online catalog of mouse strains which functions as a part of the integrated database of bioresources is available from search bars on the page of the Center ( https://brc.riken.jp ) and the Experimental Animal Division ( https://mus.brc.riken.jp/ ) websites. The BioResource Research Center also released a genomic variation database of mouse strains established in Japan and Western Europe, MoG⁺ ( https://molossinus.brc.riken.jp/mogplus/ ), and a database for phenotype-phenotype associations across the mouse phenome using data from the International Mouse Phenotyping Platform. In this review, we describe features of current version of databases related to mouse strain resources in RIKEN BioResource Research Center and discuss future views.

Keywords: Bioresource; Database; Mouse mutation information resource; Ontology; Semantic web; data integration.

Fig. 1

Basic semantic expression of RDF data model. a RDF data model is based on the simple expression in a “triple” which is a set of three entities, subject, predicate and object to describe resources in the WWW. Subjects and predicates are addressed in URIs. An object is addressed in an URI or text. b An example of RDF data model forming a knowledge graph describing a mouse mutant represented in multiple triples. Each URIs are represented with abbreviations (prefixes) listed in the box at the bottom part. Note that most URIs were defined outside RIKEN (as common vocabularies). Upper parts of the graph describe rbrc:10764 named as “C57BL/6-Shh < tm1Tshir>/Ms” with identifier “RBRC10764” has a mutation in the Shh gene of Mus musculus. The blank node is an anonymous record representing a genomic feature of C57BL/6-Shh < tm1Tshir>/Ms. (rbrc:10764). Lower part describes Shh is a homolog of the SHH gene of Homo sapiens

Fig. 2

Overview of bioresource integrated database in RIKEN BRC. Left panel shows composition of the bioresource integrated database. The integrated database for online catalog of bioresources in RIKEN BRC is composed of a back-end database and the front-end web applications. Three different applications (online catalogs of whole bioresources in RIKEN BRC, mouse and RDF data viewer and SPARQL search) are hosted by the back-end database. Right panel shows web interfaces of https://brc.riken.jp and https://knowledge.brc.riken.jp. Interface of https://mus.brc.riken.jp is shown in Fig. 2

Fig. 3

Overview of mouse strain search. a Search bars of mouse strain in RIKEN BRC are represented at the top page of Experimental Animal Division’s Page (https://mus.brc.riken.jp: upper panel) and the search page (https://mus.brc.riken.jp/en/search_for_mouse_strain: lower panel). b An example of search result for mouse strains that have homologs related to the keyword entered are listed. c Detailed information page for a mouse strain including health reports, alleles and publications. The detailed page is linked to ordering information page for mouse strains

Fig. 4

The top page of MoG+ database (https://molossinus.brc.riken.jp). Detailed explanation of this web application is available at the tutorial page (https://github.com/ttakada1/MoGplus_tutorial_2020/wiki)

Fig. 5

Overview of the strategy for phenome-wide association study in RIKEN BRC. a Overview of generation of high-quality association rules using phenotype data from IMPC. We used IMPC release 4.3 dataset contains approximately 18,000,000 data points for 2050 parameters. We made a call table of normal/abnormal phenotypes with 532 phenotypes for 3100 mutant strains (genes) by the data processing of ontology-based reduction of semantic overlaps for parameters. 3686 relationships (association rules) were obtained by the association rule mining analysis. b For evaluation and understanding of the relationships, we defined a set of PPAPs as a module of phenotypic expression for each of the 345 phenotypes. Upper panel shows a schematic diagram of PPAP in which a phenotype associated to other phenotype with statistical weight of association. Lower panel shows data representation in the web application (https://brc-riken.shinyapps.io/phenotypic_associations_across_the_mouse_phenome/) in which users can select, search and download any relationships. See Tanaka et al. (2020) [37] for details