Since 1996, the EBI has been using CORBA (Common Object Request Broker Architecture) to resolve some of the current IT problems in bioinformatics. CORBA provides a standard means to connect computational resources on different machines over the Internet. Computer applications can access these resources if they implement the programming interface to a CORBA server (specified using the Interface Definition Language - IDL). CORBA is very well suited to use in scientific visualisation and analysis applications where the integration of the multiple databases and legacy applications is important.

The GenomeBuilder is a Java tool designed to visualise and further process the EST and high-throughput genomic sequence assemblies. As an example, a user can use the tool to read in an EST cluster which is based on pair-wise comparison and then build a multiple alignment and consensus contig from the cluster using external program(s). Users can then visualise similarities between the assembled sequences and fetch additional property information for the sequences from other data sources.  The property information can then be shown as normal text or as colour coded text.

The tool uses CORBA to integrate and access the different databases and computing resources. Here is a brief summary of basic features:

• Properties of the sequences can be colour-coded or shown as a text on the sequences.
• Regions of similarity between the sequences can be highlighted.
• Linked windows mean an overall view of the assembly can be maintained in one window while working at a higher resolution in another.
• Sequences in an assembly can be edited and moved.

Currently the sequence and additional property information can be obtained from the EST cluster¹, EST², Radiation Hybrid³ and -allocation⁴ databases. New databases can be added dynamically to the GenomeBuilder if they provide an IDL and a CORBA server.

Within the Genome Builder, the AppLab⁵ program is used to generate client and server side components of external command line applications. Presently, two applications have been added: CAP3⁶ for building multiple alignments and CLEANUP⁷ to remove redundant sequences.

EST and mRNA sequence clusters can be read from the EST cluster database¹ using a browser interface shown in Figure 3. After the cluster has been read in, it is possible to retrieve additional annotations from databases and/or submit the data to external analysis tools. Figure 4 shows a multiple alignment calculated using CAP3 program⁶. In the Figure, sequences are colour coded according to a clone library and those sequences, which are derived from the same clone, are positioned on same line and connected by a thin line. The library and the clone information were retrieved from the EST database.

Figure 5 shows the graphical user interface to the CAP3 program. The interface has commands for starting and cancelling the application execution (Action menu). It is also possible to set command line parameters (Options menu) of the program as show in Figure 6. These AppLab components are composed using the information obtained from the AppLab server.

In this version (1.0.0), information about the existing databases is read from a configuration file. That information is then used in the database browser and user interface menus. New databases can be added to the configuration file if they have a CORBA server which implements the same IDLs. Currently the interfaces are made for the EST cluster, EST, RHdb and RHalloc databases

The AppLab wrapper is used to access external command line applications. The program provides tools for generating server code for the applications, based on the GCG compatible description file, and Java Bean components pertaining to the client. The client components take care of program execution and communication with the server.

The AppLab client implementation is independent of the server side implementation. However to be able to use the external applications, one needs to specify a data format which applications can understand. In the GenomeBuilder this is done by using a simple FASTA format, where the positional information of padded sequences are stored in a header along with the accession number. The FASTA data stream is then wrapped into an AppLab data event. New data formats can be added using the mime-type specification provided by the data event.

New applications can be added on the server if they use the known data format. During the program startup, the GenomeBuilder asks from the AppLab server what applications are available and builds the user interface menus accordingly.

External sites are not endorsed by EMBL-EBI