2005 Scientific Session Abstracts
Design and implementation of a data model for storage, analysis and exchange of tissue microarray data
Hye Won Lee, MS 1 ( hwon0615@snu.ac.kr ); Yu Rang Park, BS 1; Jaehyun Sim, MS 1; Woo Young Jang, MD, MS 2; Woo Ho Kim 3, MD, PhD; Ju Han Kim, MD, PhD 1,4 ; 1 Seoul National University Biomedical Informatics (SNUBI); 4Human Genome Research Institute; 3Dept. of Pathology, Seoul National University College of Medicine, Seoul 110-799, and Dept. of Pathology, Hangang Sacred Heart Hospital, Hallym University, Seoul, Republic of Korea
Context: Tissue microarray (TMA) is an array-based high-throughput technology to examine hundreds of tissue samples on a single glass slide. While DNA microarray contains thousands of probes to measure the gene expression levels of a ‘single’ specimen having the clinical (and histopathological) information, TMA contains hundreds of tissues having the clinical and histopathological information for the ‘population’. TMA has been an important tool to clinically validate the marker genes from DNA microarray analysis. To handle, exchange, and disseminate TMA data, there is a need for a standard representation of the methods used, the data generated in TMA experiments, and the clinical and histopathological information related TMA analysis. To meet the need, we propose a comprehensive data model for TMA data.
Technology: Class diagrams of Unified Modeling Language (UML) are used for representing the concepts, objects and relationships in TMA experiments. To implement the UML model as a relational database, MySQL 4.1 is used.
Design: We designed Tissue MicroArray Object Model (TMA-OM) with flexibility and extensibility. TMA-OM has flexibility that supports diverse designs of TMA experiments and extensibility that permits us to adequately describe new clinical and histopathological data elements that do not need to be predefined. TMA-OM consists of array information, experimental procedure, histopathological information and clinical information data models. Both the array information model and the experimental procedure model are created by referring to Microarray Gene Expression Object Model (MAGE-OM) that is a standard data model for DNA microarray and Minimum Information Specification for in Situ Hybridization and Immunohistochemistry Experiments (MISFISHIE) by Microarray Gene Expression Data ( MGED) society . The clinical information and the histopathological information data models are created by using Cancer Protocols established by College of American Pathologists and Common Data Elements by National Cancer Institute. The model is designed with the consideration of TMA Data Exchange Specifications developed by Association of Pathology Informatics. To implement TMA-OM as a relational database, we designed a relational schema according to formal object-to-relational-mapping rules. Applying systematic object-to-relational mapping technology supports the extensibility and flexibility of TMA data model, including the support for XML interfaces.
Results: TMA-OM contains 111 classes in 17 packages containing generalized clinical and histopathological information models. TMA-OM contains data elements for 43 major cancers but extensible to new type of cancers by generalized models.
A web-based database application, called Xperanto-TMA, was built to implement the TMA-OM. 107 tables are derived from TMA-OM. Exporting data both into an XML format that conforms to the TMA DES and into another that conforms to the DTD automatically generated from the TMA-OM are supported by Xperanto-TMA. Xperanto-TMA is available at http://xperanto.snubi.org/TMA.
Conclusions: As the number of studies with TMA is growing, there is a growing need of a data model for representing and exchanging TMA data. By developing TMA-OM, it is possible to integrate TMA data with other biological data model such as MAGE-OM promoting an understanding of the underlying biological nature.