Abstract

INTRODUCTION

Immunogenetics is dependent on a reliable and comprehensive database of variable gene segments in order to analyse the immune repertoire. Various approaches have been made to generate databases containing variable gene segments. The first and original database in this context is the Kabat database (1), which is a very valuable collection of sequences that are not necessarily included in the nucleotide sequence databases EMBL-Bank/GenBank/DDBJ. The Kabat database is the first database to classify the variable gene segments into families that are dependent on small sequence motifs. It also provides statistics on the variability of individual positions within the gene segments. The database has recently been commercialized. The next milestone was the establishment of the IMGT/LIGM database (2,3). This database collects all entries containing V gene notification from the EMBL-Bank/GenBank/DDBJ databases (4) and provides useful additional sequence annotation and classification. Furthermore, a systematic V gene nomenclature and a unique numbering system have been introduced. However, the IMGT/LIGM database does not sort the EMBL entries by their V gene sequences. In a heroic approach, the database VBASE (http://www.mrc-cpe.cam.ac.uk/vbase-ok) was compiled manually by analysing all human immunoglobulin variable gene segments known at the time. Rearrangements were assigned to a certain germ-line V gene and somatic mutations were excluded. The VBASE database is of great value although it was not updated after its first and final release in 1997.

Here we present the VBASE2 database. It follows the rationale of VBASE in sorting the EMBL entries by their V gene sequences. In contrast to VBASE, VBASE2 is generated automatically, and it provides new information and sequences as it implements the current knowledge derived from the genome sequencing projects by linking to the Ensembl Genome Browser (5). VBASE2 also connects the existing immunoglobulin sequence databases, thereby integrating the distinct knowledge resources.

THE VBASE2 DATASET

The current VBASE2 dataset contains immunoglobulin germ-line V genes from the heavy chain and lambda and kappa light chain loci of human and mouse. The current release holds 498 human and 554 mouse V gene sequences.

Automatic generation

The sequence data and database cross-references provided by VBASE2 are generated automatically so that manual annotation is not required. An overview about the procedure is given in Figure ​Figure1.1. By a BLAST search (6) of known germ-line V genes all potential V gene sequences are extracted from the EMBL-Bank, including the high throughput genomic (HTG) and whole genome shotgun (WGS) sections (4). Potential V gene sequences from Ensembl are extracted by a BLAST search against the Ensembl chromosome sequences. The DNAPLOT software is used to align, sort and compare the V gene sequences, identify J elements, RSS elements and pseudogenes. Synthetic sequences are detected and removed. All germ-line configured V genes are matched to the rearranged sequences. To assign a rearrangement to a germ-line sequence a 100% match in the V gene region is required. Thus, the sequence comparison is restricted to the FR1–FR3 region, excluding potential N nucleotides in CDR3. The current procedure assigns V gene alleles to different V gene entries, and allele assignment is not yet included in the database. V gene families are assigned using family consensus sequences. In addition, DNAPLOT is used to compare the VBASE2 dataset with the LIGM dataset from the IMGT database, the VBASE database and the last freely available version of the Kabat database (ftp://ftp.ebi.ac.uk/pub/databases/kabat/). Owing to the changes in the source sequence databases, Ensembl and EMBL-Bank/GenBank/DDBJ, the VBASE2 dataset is updated regularly.

Open in a separate window

Figure 1

The data generation procedure. The procedure analysing the V gene sequences retrieved by the BLAST search is performed using the DNAPLOT program and interconnecting Perl scripts. EMBL-Bank entries containing a single V gene are filtered for synthetic sequences. All V gene sequences are checked for stop codons to detect pseudo genes. Rearrangements are detected by an alignment against J elements, RSS element detection allows the detection of germ-line configured V gene entries. In a multiple alignment step, all rearranged V gene sequences are matched to the germ-line configured V genes. All germ-line V genes are matched against the VBASE, IMGT/LIGM and KABAT database.

Cross-references, V gene annotation and features

Each V gene entry holds a list of source references linking to EMBL-Bank and/or Ensembl (Figure ​(Figure2).2). If the EMBL-Bank reference is a BAC sequence, the V gene position within the BAC is given, as many BAC sequences have not yet been annotated. Sequences containing stop codons are labelled as pseudogenes, V genes allocated to another chromosomal locus are marked as orphans. As several names may have been assigned to the same V gene all known names for each V gene are listed. Furthermore, hits in the IMGT-, KABAT- and VBASE-databases are shown. These cross-references allow access to manually annotated data available in these databases. Also, the protein translation and the positions of the complementary determining regions (CDRs) are indicated.

Open in a separate window

Figure 2

V gene entry example. The V gene entry page is divided into five sections: general information about the V gene, the source sequences from which the entry was created, cross-references to other immunological databases, sequence features and the nucleotide sequence.

ACCESSING THE VBASE2 DATABASE

The VBASE2 database can be accessed at http://www.vbase2.org. V gene entries can be requested either by a text-based query or a sequence similarity search with the DNAPLOT tool.

The DNAPLOT query

To compare a complete V gene sequence or rearrangement with the VBASE2 dataset, the DNAPLOT query provides a sequence similarity search tool. The query returns a V gene alignment referring to the IMGT unique numbering (3), containing the query sequence and the best VBASE2 matches. Queries containing a V gene rearrangement return the name of the D- and J-element and also the automatically assigned V gene family is given (Figure ​(Figure33).

Open in a separate window

Figure 3

The DNAPLOT query output. The figure shows the 5′-part of the V gene alignment, the alignment of D- and J-elements and the translation of the junction of the query sequence ‘No_Name’.

IMPLEMENTATION

VBASE2 is implemented in a relational database structure using PostgreSQL DBMS. The web interface uses PHP scripts for dynamic web pages. The website requires a HTML 4.0-compliant browser with JavaScript enabled. The automatic generation procedure uses the NCBI BLASTALL program, the DNAPLOT program and Perl scripts.

CONCLUSIONS

VBASE2 connects several separated data collections and thereby combines all V gene annotation and classification data from the distinct resources. Furthermore, it shows the chromosomal location of a V gene in Ensembl, and a DAS server enables the display of the V genes in the Ensembl Genome Browser. During the automatic data generation process, sequences are sorted and evaluated only on the basis of their sequence information. Classification and cross-references allow the user to validate the sequence quality. Currently, the VBASE2 database contains germ-line V gene sequences of the immunoglobulin loci of human and mouse. A forthcoming challenge in the future development of the database is the assignment of haplotypes and V gene alleles. Another important step is the extension of the stored V gene sequence to the end of the RSS element. Furthermore, the scope of the database will be extended; as the process of sequence extraction and evaluation only requires the extension of the computer programs and the underlying sequence tables, the database can be expanded to T-cell receptor sequences and to other species.

ACKNOWLEDGEMENTS

REFERENCES