ABSTRACT
IMGT, the international ImMunoGeneTics database, is an integrated database
specializing in immunoglobulins, T-cell receptors (TcR) and major histocompatibility complex (MHC) of all
vertebrate species, initiated and co-ordinated by Marie-Paule Lefranc, CNRS, Montpellier II University, Montpellier, France
(lefranco{at}ligm.crbm.cnrs-mop.fr). IMGT includes two databases: LIGM-DB (for immunoglobulins and TcR) and MHC/HLA-DB. IMGT comprises expertly annotated sequences and alignment
tables. LIGM-DB contains more than 19 000 immunoglobulin and TcR sequences from 78
species. MHC/HLA-DB contains class I and class II human leukocyte antigen alignment tables.
An IMGT tool, DNAPLOT, developed for immunoglobulins, TcR and MHC sequence
alignments, is also available. IMGT works in close collaboration with the EMBL
database. IMGT goals are to establish a common data access to all
immunogenetics data, including sequences, oligonucleotide primers, gene maps
and other genetic data of immunoglobulins, TcR and MHC molecules, and to
provide a graphical user-friendly data access. IMGT will have important implications in medical
research (repertoire in autoimmune diseases, AIDS, leukemias, lymphomas),
therapeutical approaches (antibody engineering), genome diversity and genome
evolution studies. IMGT can be accessed at http://imgt.cnusc.fr:8104 and
http://www.ebi.ac.uk/IMGT
The molecular synthesis of the immunoglobulin and T-cell receptor (TcR) chains (
1
,
2
) is particularly complex and unique as it includes biological mechanisms such
as DNA molecular rearrangements in seven loci (three for immunoglobulins and
four for TcR) located on four different chromosomes in human, nucleotide
deletions and insertions at the rearrangement junctions, and hypermutations in
the immunoglobulin loci. The number of potential protein forms of
immunoglobulins and TcR is almost unlimited. Owing to the complexity and high
number of published sequences, data control and detailed annotations are a very
difficult task for the generalist databanks: EMBL (
3
), GenBank (
4
), DDBJ. Furthermore, until now, only poor efforts have been made to standardize
the description of the immunoglobulin and TcR sequences at the nucleotide or
protein level. Only few feature labels are specifically used in generalist
databases for immunoglobulin and TcR annotations (seven in EMBL) and this often
leads to errors or misinterpretations. These observations together with the
proposal made by Fuchs and Cameron (
5
) to create specialized databases in collaboration with the generalist
databases, were the starting point of IMGT in 1992 (
6
) (see Fig.
1
for the IMGT home page at http://imgt.cnusc.fr:8104). Before the physical
implementation of the database, the main and the longest objective was to
establish rules for describing immunoglobulin and TcR sequences of any species.
This was the major foundation for a consistent expertise.
IMGT keywords for immunoglobulins and TcR comprise the following. (i)
General keywords
. Indispensable for the sequence assignments, they are described in an
exhaustive and non-redundant list, and are organized in a tree structure. (ii)
Specific keywords
. They are more specifically associated with particularities of the sequences
(orphon, pseudogene...) or to diseases (leukemia, lymphoma, tumor...). The list
is not definitive and new specific keywords can easily be added if needed.
The whole list of keywords can be reached using WWW browser at the URL
http://imgt.cnusc.fr:8104/textes/LECT/kw.html.
Immunoglobulin and TcR sequences have been analyzed at the DNA and protein level
in order to define a list of labels for the structural and functional motifs.
More than 160 labels were shown to be necessary for an accurate annotation. The
annotation is the most critical step and a very time-consuming process as about 50 sequences a week can be annotated by an
experienced annotator. Levels of annotation have been defined, which allow the
users to query sequences in IMGT/LIGM-DB even though they are not fully annotated. The list of labels with their
corresponding definition and main schemas are available at the URL:
http://imgt.cnusc.fr:8104/textes/LECT/labeldef.html (Fig.
2
).
LIGM-DB development is mainly based on a relational model organization. The
database is maintained with SYBASE as relational DBSM (Data Base System
Manager) on Unix IBM workstation at CNUSC (Centre National Universitaire Sud de
Calcul) in Montpellier (France). CNUSC is in charge of the computing
exploitation. New releases of the relational schema and updates of the database
structure, that closely follow the results of biological research, are under
LIGM and CNUSC responsibility.
In November 1996, LIGM-DB contained 19 540 nucleic acid sequences of immunoglobulins or TcR from
78 species. IMGT sequences are identified by the EMBL accession number. IMGT
data comprise core data that consist of sequence data, bibliographical
references, taxonomic data retrieved from EMBL entries, completed with
annotations, specific analysis and expertise provided by LIGM. IMGT/LIGM-DB standardized keywords have been assigned to all entries, and 7908
sequences are now fully annotated. Since August 1996, the IMGT/LIGM-DB content follows closely the immunoglobulin and TcR EMBL one, with the
advantage of being deleted from sequences which have previously been wrongly
assigned to immunoglobulins and TcR.
The unique source of data is the generalist database EMBL. Once the sequences
are allowed by the authors to be made public, EMBL sends automatically
immunoglobulin and TcR sequences to LIGM by mail. After control by LIGM
curators, sequences are scanned in order to store IMGT non-specific information, such as bibliographical references and taxonomic
data.
Standardized keywords are so far assigned manually to each new sequence by LIGM
annotators. Procedure for the automation of the IMGT keyword attribution is in
development.
The annotation of sequences is the most limiting step in the expertise of the data.
Several approaches have been developed in order to increase the number of
annotated sequences per month, and efforts are currently done to improve LIGM
efficiency in this field.
Automatic motif recognition
. The C written general algorithm for motif searches, BioMotif, developed by the
Laboratoire de Physique Mathématique of Montpellier, France, has been specifically adapted for
immunoglobulin and TcR sequences. This algorithm, designated as LIGMotif and
based on the use of EMBL flat files, scans the nucleic acid sequence for
immunoglobulin or TcR specific motifs (characteristic amino acids in conserved
positions...), according to the presence of information such as receptor and
chain type. At the end of the search, it provides a text file which contains
potential solutions for delimitation of functional or structural subregions. It
also provides the FR (FRamework) and CDR (Complementarity Determining Region)
delimitations (
1
).
Annotations in delayed conditions
. In order to make the annotators independent from Internet connection and allow
them to annotate `in any place', we have developed a simple text mode release
of the annotation module that facilitates the data acquisition on any local
computer. Resulting annotations are then sent by mail, ftp or tape to LIGM.
Annotators can also use text files resulting from LIGMotif analysis and, after
control of the annotations, include them into IMGT.
A tool for immunoglobulin, TcR and MHC sequence alignments: DNAPLOT
. Immunologists mainly use sequence comparison either to search similar or
identical sequences in databases, or to classify immunoglobulins and TcR
sequences in subgroups, in which the sequences share more than 75% similarity
and can be detected by the same probe. The Institut Für Genetic (IFG) of Köln, Germany, has developed a program DNAPLOT which generates,
displays and analyzes nucleotide sequence alignments. DNAPLOT is complementary
to existing programs, such as GDE, CLUSTALW, FASTA, BLAST or READSEQ, and does
not replace their functions. It can also propose assignment of rearranged or
expressed variable genes to the potential germline genes. DNAPLOT is available
at:
http://www.genetik.uni-koeln.de/dnaplot/
and from the IMGT Home page.
No restrictions are placed on the use or redistribution of the IMGT data.
Currently, IMGT is available through Internet and on the quaterly CD-ROM distributed by the EMBL data library.
Flat files are produced in collaboration with EBI. Names of entries remain the
EMBL accession number. IMGT/LIGM-DB flat file typical entries provide LIGM expertise: standardized LIGM
keywords appear in KW code lines, complement to definition in DE lines and
sequence description with LIGM labels in FT code lines. Core data, as well as
cross-references to other databases in DR lines are kept from EMBL. Flat file
format allows IMGT/LIGM-DB data to be compatible with the most efficient software for information
retrieval, data manipulation such as the largely distributed browser SRS, which
also allows consultation of the cross-referenced databases (available at http://www.ebi.ac.uk/srs/srsc).
IMGT/LIGM-DB flat files are available on EMBL anonymous ftp server (ftp.ebi.ac.uk in
pub/databases/imgt) and are also distributed with many other databases on the
EMBL CD-ROM.
A WWW IMGT server has been installed at CNUSC and can be reached with Mosaic and
Netscape WWW browser at the URL http://imgt.cnusc.fr:8104. The biologist needs
were taken into account for the development of the interface WWW-SYBASE which allows users to create very specific and structured queries
combining aspects of relational database and hypertext. Requests can be
performed through distinct modules that allow to classify search criteria type.
At the issue of a run, a number of resulting sequences is proposed and it is
then possible to either look at the solutions, or to add new conditions to
modify the results, keeping in memory the previously selected criteria. There
are several ways to retrieve the results, in particular it is possible to
extract specific coding regions from the query resulting sequences even though
alignment tools are not yet integrated into IMGT (Fig.
4
). Links with Medline are now available.
IMGT is developed by LIGM (Montpellier, France) in collaboration with CNUSC
(Montpellier, France), EMBL-EBI (Hinxton, UK), ICRF (London, UK), IFG (Köln, Germany), BPRC (Rijswijk, The Netherlands) and EUROGENTEC S.A.
(Seraing, Belgium). The information provided by IMGT is of much value to
clinicians and biological scientists in general. The main objectives for the
next 3 years include the development of a WWW interface for direct submission
of the data by the authors, development of MHC/HLA-DB and extension to all species. New specific databases will be developed
and integrated into IMGT: a protein database for immunoglobulins and TcR which
will contain translations of potentially functional and ORF sequences from LIGM-DB, and protein data from Kabat (
7
) and SWISS_PROT (
8
), and an oligonucleotide primer database for immunoglobulins, TcR and MHC. IMGT
will include, in the future, analysis of genetics data and displays of physical
maps. IMGT is designed to allow common access to all immunogenetics data. This
approach is based on a very tight collaboration with EMBL for the nucleotide
sequence data, with SWISS-PROT for the protein sequence data and with IGD for providing a user
friendly interface for the mapping and genetic data. Particular attention will
be given to the establisment of cross-referencing links to other databases pertinent to the users of IMGT.
CNUSC WWW server at http://imgt.cnusc.fr:8104. Contact Denys.Chaume{at}cnusc.fr.
EBI servers at http://www.ebi.ac.uk/imgt;, ftp.ebi.ac.uk
(folder/pub/databases/imgt); contact malik@ebi.ac.uk
For comments and suggestions contact giudi@ligm.crbm.cnrs-mop.fr
IMGT initiator and coordinator: Marie-Paule Lefranc, Laboratoire d'ImmunoGénétique Moléculaire, LIGM, UMR CNRS 5535, BP5051, 1919 route de Mende, 34033
Montpellier Cedex 1, France; Tel: +33 467 61 36 34; Fax: +33 467 04 02 31; Email: lefranco{at}ligm.crbm.cnrs-mop.fr
We thank Gérard Mennessier for the development of LIGMotif. We are deeply grateful
to Valérie Barbié, Anne Bouisson, Géraldine Folch, Sophie Lefebvre, Nathalie Pallares and Gaëlle Rousseaux who are the present LIGM-DB annotators. IMGT is funded by the
European Union's BIOMED1 and BIOTECH programmes, the CNRS (Centre National de
la Recherche Scientifique), and the MENESR (Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche). Subventions have been received from
Association pour la Recherche sur le Cancer, Association de Recherche sur la
Polyarthrite, Fondation pour la Recherche Médicale, Groupement de Recherche et d'Etude sur les Génomes and the Région Languedoc-Roussillon.
*To whom correspondence should be addressed. Tel: +33 467 61 36 34; Fax: +33 467
04 02 31; Email: lefranco{at}ligm.crbm.cnrs-mop.fr
REFERENCES
Return