BODB Conceptual Overview

Our current version of BODB combines features of an SDB (Summary Database) and Model Repository (as in Brain Models on the Web [BMW], Bischoff-Grethe, Spoelstra & Arbib, 2001), supplemented with a variety of new features. The system provides data entities for storing article information, brain operating principles, models, generic summary data, brain-imaging experimental data, and, finally , their relations.

Bodb overview

The entity relationship schema of BODB’s fundamental entities.

Article Information: This is quite standard, functioning as a bibliography, keeping information on publications cited in other entries in BODB. As usual, each item is categorized into one of 7 types: journal, book, chapter, thesis, conference, electronic, and unpublished. Brain Operating Principle (BOP): This originated from the observation that key concepts concerning brain mechanisms are sometimes implicit in many references but integrated in none. The Brain Operating Principle entity is designed to serve as a repository for maintaining structured and succinct concepts about “how the brain (or a key – functional or structural – subsystem) works”, whether extracted and generalized from empirical data, or exemplified in computational models. We designed this entity with the m:n relationship linking to Article information since one brain operating principle could be derived from many articles, while one article could be a source of information for two or more principles.

Model: This is designed to function as a repository for descriptions of computational models of brain mechanisms with fields available to provide linkages to actual implementations, simulations, documentation and descriptions. We designed this entity, too, with the m:n relationship to Article information. Another m:n relationship links the Model and Brain Operating Principle entities because one model could be derived from many principles, while one principle could be exemplified in many models. Since one model could be synthesized from other models, the m:n self-relationship is also supplied to this entity.

Brain Region: This entity contains names and hierarchies of regions of the human brain, currently based on data from the Research Imaging Center, The University of Texas Health Science Center at San Antonio. These brain region names are used to categorize entries related to specific human brain regions, enhancing entry-search capability.

Summaries of Empirical Data and Simulation Results: BODB provides forms for two types of summaries: one for summarizing experimental data (SED) and the other for summarizing simulation results (SSR). For SEDs, BODB currently provides generic forms to contribute to a repository for summarizing related facts, experimental settings and results from one or several publications. A specific form for brain imaging experimental data (see below) is also provided. For SSRs, BODB provides only a generic form for use in documenting each related set of key findings associated with a model. Each SED is amenable for use to support the design of a model or to be used in testing a model – whose SSRs may explain the SED or be contradicted by SEDs. The latter situation may either be used to downgrade confidence in the model or to point the way to further efforts that build on the current model.\

Brain Imaging Experimental Data: Currently, this entity provides the one specifically structured type of Summary of Experimental Data for BODB. It serves primarily as a repository for tables of brain-imaging experimental data. We adapted the Talairach Daemon (TD) developed by the San Antonio group as well as ideas from USCBP’s own NeuARt project (Dashti et al., 2001) to enable the user to graphically compare and contrast the experimental data on brain-slice images. One brain imaging article could have two or more experiment sets, and almost every brain imaging experiment will relate to multiple brain regions.

In addition to the basic entities described so far, we added a set of tables that are responsible for maintaining block diagrams for models and anatomical relationships, etc., with supplementary information such as box area (coordinates) and box name. These diagrams not only help the user visualize Model entries, but also provide connections between Models via hierarchical relations.

Connections among entities not only allow the user to go back and forth between entries via their relations, but provide options for cross search as well (e.g. to search models by related brain operating principles).