The NIF DISCO Framework: facilitating automated integration of neuroscience content on the web

Abstract

This paper describes the capabilities of DISCO, an extensible approach that supports integrative Web-based information dissemination. DISCO is a component of the Neuroscience Information Framework (NIF), an NIH Neuroscience Blueprint initiative that facilitates integrated access to diverse neuroscience resources via the Internet. DISCO facilitates the automated maintenance of several distinct capabilities using a collection of files 1) that are maintained locally by the developers of participating neuroscience resources and 2) that are "harvested" on a regular basis by a central DISCO server. This approach allows central NIF capabilities to be updated as each resource's content changes over time. DISCO currently supports the following capabilities: 1) resource descriptions, 2) "LinkOut" to a resource's data items from NCBI Entrez resources such as PubMed, 3) Web-based interoperation with a resource, 4) sharing a resource's lexicon and ontology, 5) sharing a resource's database schema, and 6) participation by the resource in neuroscience-related RSS news dissemination. The developers of a resource are free to choose which DISCO capabilities their resource will participate in. Although DISCO is used by NIF to facilitate neuroscience data integration, its capabilities have general applicability to other areas of research.

Figure 1

A schematic overview of the DISCO operational framework. (1) The DISCO server updates its registry of participating Web resources from the NIF Registry. (2) All resources are checked for their DISCO files. (3) When found, each file is processed by its corresponding DISCO capability engine, and its content is synchronized with the associated NIF component. Regular users of NIF and the Entrez systems use the information provided by the DISCO server to find the information from the collection of participating Web resources. (In addition to supporting this operational framework, DISCO also provides tools to allow resource staff to create and update their DISCO files, and to help NIF staff coordinate the various DISCO-enabled activities.)

Figure 2

The DISCO main file for NeuronDB (http://senselab.med.yale.edu/neurondb/disco.xml) contains several “service” nodes, corresponding to the different DISCO capabilities in which NeuronDB is participating. Each node has a type describing the capability, a format and a URL pointer to the DISCO XML file corresponding to that specific capability. A single type of DISCO capability can be described in more than one format (as shown in this figure for the “Resource Description” and “Interoperation” service nodes).

Figure 3

The resource description authoring tool on the DISCO server that allows a resource developer to create and edit the DISCO resource description for a resource.

Figure 4

This figure shows how the NCBI Entrez system uses LinkOut information provided by three neuroscience resources via the NIF DISCO LinkOut capability. The left screen shows Entrez’s “LinkOut - more resources” information (expanded) for a PubMed paper. Down at the bottom of that page, under “Miscellaneous”, are three links provided by the NIF. To the right are three data-pages (from ModelDB, NeuronDB, and Neuromorpho.org respectively) generated by clicking on those URLs, all of which relate to the PubMed paper.

Figure 5

This figure shows a portion of CCDB’s DISCO LinkOut XML file. The first “oid” node shows how a specific PubMed ID (15988042) is linked to a specific CCDB URL (http://ccdb.ucsd.edu/sand/main?event=displayAllProjectProds&mpid=48&ptype=sproject).

Figure 6

This screen shows a portion of the DISCO Web Interoperation file created to extract neurological disorder information from the National Institute of Neurological Disorders and Stroke Web site. The format is XML-based and includes information about interfaces and parameters (analogous to database tables and fields) among other elements to be passed, and the necessary logic to navigate the Web site to locate, extract, and transform the information into relational tables.

Figure 7

The DISCO Web Interoperability Control Panel. This tool executes “disco.interop” scripts. The figure shows the URL of the National Institute of Neurological Disorders and Stroke (NINDS) DISCO interoperability script that collects disorder information from the site’s Web pages. The NINDS disorder information is scattered among several pages. During the extraction process, this tool first extracts disorder URLs from index pages. Later it queries each of those URLs to extract the disorder information. This process is recursive into linked pages containing related information about the disorder to create other tables. All information collected by this script is passed to NIF servers where it is stored along with similar information from other NIF resources.

Figure 8

Data Federation results in the NIF’s user search interface after searching for “ataxia”. On the left, the “Disease (24)” category was expanded, and the “NINDS: ninds (24)” source was clicked. The first disease “ataxia” is presented in a list of a total of 24 diseases. Notice the relation between the columns shown in this figure (Name, What_is, Treatment and Prognosis) with “Disorder Detail” fields shown in figure 7.

Figure 9

This figure shows a portion of the DISCO Terminology data (in XML format) describing the terms used in the Olfactory Receptors Database (ORDB), http://senselab.med.yale.edu/ordb.

Figure 10

This figure shows a tool designed to search the DISCO Terminology files of multiple NDG resources. In this example, the search string “%cerebellar%” (the percent sign is used as a wildcard) has been entered and all matching terms are returned from multiple resources, together with information describing each term

Figure 11

This figure shows the DISCO Dashboard Control Panel. For each participating resource, this panel indicates which DISCO capabilities are being utilized. In addition the Dashboard provides controls allowing NIF staff who are logged in to import DISCO files provided by the participating resources, as well as tools to export and synchronize that information with other NIF components.