Genome Database 
Johns Hopkins University 
2024 E. Monument Street 
Baltimore, MD 21205-2236 

Stanley Letovsky 
Informatics Director 

Robert Cottingham 
Operations Director 

Telephone for both: 410/955-9705 
Fax for both: 410/614-0434 

David Kingsbury 
Director, 1993­97* 
 
*Now at Chiron Pharmaceuticals, Emeryville, California 
 
 
 
 
 
 

In lieu of individual abstracts, research projects and investigators at GDB are represented in this narrative. More information can be found on GDB's Web site . 
 

The release of Version 6 of the Genome Database (GDB) in January 1996 signaled a major change for both the scientific community and GDB staff. GDB 6.0 introduced a number of significant improvements over previous versions of GDB, most notably a revised data representation for genes and genomic maps and a new curatorial model for the database. These new features, along with a remodeled database structure and new schema and user interface, provide a resource with the potential to integrate all scientific information currently available on human genomics. GDB rapidly is becoming the international biomedical research community's central source for information about genomic structure, content, diversity, and evolution. 

A New Data Model 

Inherent in the underlying organization of information in GDB is an improved model for genes, maps, and other classes of data. In particular, genomic segments (any named region of the genome) and maps are being expanded regularly. New segment types have been added to support the integration of mapping and sequencing data (for example, gene elements and repeats) and the construction of comparative maps (syntenic regions). New map types include comparative maps for representing conserved syntenies between species and comprehensive maps that combine data from all the various submitted maps within GDB to provide a single integrated view of the genome. Experimental observations such as order, size, distance, and chimerism are also available. 

Through the World Wide Web, GDB links its stored data with many other biological resources on the Internet. GDB's External Link category is a growing collection of cross-references established between GDB entities and related information in other databases. By providing a place for these cross-references, GDB can serve as a central point of inquiry into technical data regarding human genomics. 

Direct Community Data Submission and Curation 

Two methods for data submission are in use. For individuals submitting small amounts of data, interactive editing of the database through the Web became available in April 1996, and the process has undergone several simplifications since that time. This continues to be an area of development for GDB because all editing must take place at the Baltimore site, and Internet connections from outside North America may be too slow for interactive editing to be practical. Until these difficulties are resolved, GDB encourages scientists with limited connectivity to Baltimore to submit their data via more traditional means (e-mail, fax, mail, phone) or to prepare electronic submissions for entry by the data group on site. 

For centers submitting large quantities of data, GDB developed an electronic data submission (EDS) tool, which provides the means to specify login password validation and commands for inserting and updating data in GDB. The EDS syntax includes a mechanism for relating a center's local naming conventions to GDB objects. Data submitted to GDB may be stored privately for up to 6 months before it automatically becomes public. The database is programmed to enforce this Human Genome Project policy. Detailed specifications of GDB's EDS syntax and other submission instructions are available (EDS prototype). 

Since the EDS system was implemented, GDB has put forth an aggressive effort to increase the amount of data stored in the database. Consequently, the database has grown tremendously. During 1996 it grew from 1.8 to 6.7 gigabytes. 

To provide accountability regarding data quality, the shift to community curation introduced the idea that individuals and laboratories own the data they submit to GDB and that other researchers cannot modify it. However, others should be able to add information and comments, so an additional feature is the community's ability to conduct electronic online public discussions by annotating the database submissions of fellow researchers. GDB is the first database of its kind to offer this feature, and the number of third-party annotations is increasing in the form of editorial commentary, links to literature citations, and links to other databases external to GDB. These links are an important part of the curatorial process because they make other data collections available to GDB users in an appropriate context. 

Improved Map Representation and Querying 

Accompanying the release of GDB 6.0, the program Mapview creates graphical displays of maps. Mapview was developed at GDB to display a number of map types (cytogenetic, radiation hybrid, contig, and linkage) using common graphical conventions found in the literature. Mapview is designed to stand alone or to be used in conjunction with a Web browser such as Netscape, thereby creating an interactive graphical display system. When used with Netscape, Mapview allows the user to retrieve details about any displayed map object. 

Maps are accessed through the query form for genomic segment and its subclasses via a special program that allows the user to select whole maps or slices of maps from specific regions of interest and to query by map type. The ability to browse maps stored in GDB or download them in the background was also incorporated into GDB 6.0. 

GDB stores many maps of each chromosome, generated by a variety of mapping methods. Users who are interested in a region, such as the neighborhood of a gene or marker, will be able to see all maps that have data in that region, whether or not they contain the desired marker. To support database querying by region of interest, integrated maps have been developed that combine data from all the maps for each chromosome. These are called Comprehensive Maps. 

Queries for all loci in a region of interest are processed against the comprehensive maps, thereby searching all relevant maps. 

Comprehensive maps are also useful for display purposes because they organize the content of a region by class of locus (e.g., gene, marker, clone) rather than by data source. This approach yields a much less complex presentation than an alignment of numerous primary maps. Because such information as detailed orders, order discrepancies between maps, and nonlinear metric relations between maps is not always captured in the comprehensive maps, GDB continues to provide access to aligned displays of primary maps. 

A Variety of Searching Strategies 
Recognizing the eclectic user community's need to search data and formulate queries, GDB offers a spectrum of simple to complex search strategies. In addition, direct programming access is available using either GDB's object query language to the Object Broker software layer or standard query language to the underlying Sybase relational database. 

Querying by Object Directly from GDB's Home Page 
The simplest methods search for objects according to known GDB accession numbers; sequence database­accession numbers; specified names, including wildcard symbols that will automatically match synonyms and primary names; and keywords contained anywhere in the text. 

Querying by Region of Interest 
A region of interest can be specified using a pair of flanking markers, which can be cytogenetic bands, genes, amplimers (sequence tagged sites), or any other mapped objects. Given a region of interest, the comprehensive maps are searched to find all loci that fall within them. These loci can be displayed in a table, graphically as a slice through a comprehensive map, or as slices through a chosen set of primary maps. A comprehensive map slice shows all loci in the region, including genes, expressed sequence tags (ESTs), amplimers, and clones. A region also can be specified as a neighborhood around a single marker of interest. 

Results of queries for genes, amplimers, ESTs, or clones can be displayed on a GDB comprehensive map. Results are spread across several chromosomes displayed in Mapview (see figure below). A query for all the PAX genes (specified as symbol = PAX* on the gene query form) retrieves genes on multiple chromosomes. Double-clicking on one of these genes brings up detailed gene information via the Web browser. 

Querying by Polymorphism 
GDB contains a large number of polymorphisms associated with genes and other markers. Queries can be constructed for a particular type of marker (e.g., gene, amplimer, clone), polymorphism (i.e., dinucleotide repeat), or level of heterozygosity. These queries can be combined with positional queries to find, for example, polymorphic amplimers in a region bounded by flanking markers or in a particular chromosomal band. If desired, the retrieved markers can be viewed on a comprehensive map. 

Work in Progress 

Mapview 2.3 
Mapview 2.1, the next generation of the GDB map viewer, was released in March 1997. The latest version, Mapview 2.3, is available in all common computing environments because it is written in the Java programming language. Most important, the new viewer can display multiple aligned maps side by side in the window, with alignment lines indicating common markers in neighboring maps. As before, users can select individual markers to retrieve more information about them from the database. 

GDB developers have entered into a collaborative relationship with other members of the bioWidget Consortium so the Java-based alignment viewer will become part of a collection of freely available software tools for displaying biological data. 

Future plans for Mapview include providing or enhancing the ability to generate manuscript-ready Postscript map images, highlight or modify the display of particular classes of map objects based on attribute values, and requery for additional information. 

Variation 
Since its inception, GDB has been a repository for polymorphism data, with more than 18,000 polymorphisms now in GDB. A collaboration has been initiated with the Human Gene Mutation Database (HGMD) based in Cardiff, Wales, and headed by David Cooper and Michael Krawczak. HGMD's extensive collection of human mutation data, covering many disease-causing loci, includes sequence-level mutation characterizations. This data set will be included in GDB and updated from HGMD on an ongoing basis. The HGMD team also will provide advice on GDB's representation of genetic variation, which is being enhanced to model mutations and polymorphisms at the sequence level. These modifications will allow GDB to act as a repository for single-nucleotide polymorphisms, which are expected to be a major source of information on human genetic variation in the near future. 

Mouse Synteny 
Genomic relationships between mouse and man provide important clues regarding gene location, phenotype, and function (see figures at left). One of GDB's goals is to enable direct comparisons between these two organisms, in collaboration with the Mouse Genome Database at Jackson Laboratory. GDB is making additions to its schema to represent this information so that it can be displayed graphically with Mapview. In addition, algorithmic work is under way to use mapping data to automatically identify regions of conserved synteny between mouse and man. These algorithms will allow the synteny maps to be updated regularly. An important application of comparative mapping is the ability to predict the existence and location of unknown human homologs of known, mapped mouse genes. A set of such predictions is available in a report at the GDB Web site, and similar data will be available in the database itself in the spring of 1998. 

Collaborations 
GDB is a participant in the Genome Annotation Consortium (GAC) project, whose goal is to produce high-quality, automatic annotation of genomic sequences. Currently, GDB is developing a prototype mechanism to transition from GDB's Mapview display to the GAC sequence-level browser over common genome regions. GAC also will establish a human genome reference sequence that will be the base against which GDB will refer all polymorphisms and mutations. Ultimately, every genomic object in GDB should be related to an appropriate region of the reference sequence. 

Sequencing Progress 
The sequencing status of genomic regions now can be recorded in GDB. Based on submissions to sequence databases, GAC will determine genomic regions that have been completed. GDB also will be collaborating with the European Bioinformatics Institute, in conjunction with the international Human Genome Organisation (HUGO), to maintain a single shared Human Sequence Index that will record commitments and status for sequencing clones or regions. As a result, the sequencing status of any region can be displayed alongside other GDB mapping data. 

Outreach 

The Genome Database continues to seek direct community feedback and interact with the broader science community via various sources: 

• International Scientific Advisory Committee meets annually to offer input and advice. 
• Quarterly Review Committee confers frequently with the staff to track GDB progress and suggest change. 
• HUGO nomenclature, chromosome, and other editorial committees have specialized functions within GDB, providing official names and consensus maps and ensuring the high quality of GDB's content. 
Copies of GDB are available worldwide from ten mirror sites (nodes) that make the data more easily accessible to the international research community. GDB staff meet annually with node managers to facilitate interaction and to benefit from other user perspectives.