MOTIVATION: Expressed Sequence Tags (ESTs) are cheap, easy and quick to obtain relative to full genomic sequencing and currently sample more eukaryotic genes than any other data source. They are particularly useful for developing Sequence Tag Sites (STSs for mapping), polymorphism discovery, disease gene hunting, mass spectrometer proteomics, and most ironically for finding genes and predicting gene structure after the great effort of genomic sequencing. However, ESTs have many problems and the public EST databases contain all the errors and high redundancy intrinsic to the submitted data so it is often found that derived database views, which reduce both errors and redundancy, are more effective starting points for research than the original raw submissions. Existing derived views such as EST cluster databases and consensus databases have never published supporting evidence or intermediary results leading to difficulties trusting, correcting, and customizing the final published database. These difficulties have lead many groups to wastefully repeat the complex intermediary work of others in order to offer slightly different final views. A better approach might be to discover the most expensive common calculations used by all the approaches and then publish all intermediary results. Given a globally accessible database with a suitable component interface, like the JESAM software described in this paper, the creation of customized EST-derived databases could be achieved with minimum effort. RESULTS: Databases of EST and full-length mRNA sequences for four model organisms have been self-compared by searching for overlaps consistent with contiguity. The sequence comparisons are performed in parallel using a PVM process farm and previous results are stored to allow incremental updates with minimal effort. The overlap databases have been published with CORBA interfaces to enable flexible global access as demonstrated by example Java applet browsers. Simple cDNA supercluster databases built as alignment database clients are themselves published via CORBA interfaces browsable with prototypical applets. A comparison with UniGene Mouse and Rat databases revealed undesirable features in both and the advantages of contrasting perspectives on complex data. AVAILABILITY: The software is packaged as two Jar files available from: URL: http://corba.ebi.ac.uk/EST/jesam/jesam. html. One jar contains all the Java source code, and the other contains all the C, C++ and IDL code. Links to working examples of the alignment and cluster viewers (if remote firewall permits) can be found at http://corba.ebi.ac.uk/EST. All the Washington University mouse EST traces are available for browsing at the same URL.

DNA sequence clustering is an effective aid of the comprehension, summarization and compression of DNA sequence databases. Previous work created programs suitable for the comparison and clustering of cDNA sequences but new enhanced programs have been written to cluster genomic DNA fragments, large EST projects, and entire DNA databases. Three new programs (ICAtools) are discussed: ICAass, N2tool, and ICAmatches. ICAass has been used to compress the EMBL database by hiding or removing sequences with various degrees of redundancy. It also has the fastest database querying mode. N2tool provides fast and sensitive clustering of genomic fragment databases on the basis of small areas of local similarity. N2tool has proven utility in the discovery of contaminating vector or other artefactual sequence when the potential contaminant is not otherwise known. ICAmatches is a new cluster analysis program that uses a novel alignment style to present multiple alignment summaries. All the tools are convenient to use because they share a common memory-frugal index format and accept most DNA sequence formats directly.

Miropeats displays DNA sequence similarity information graphically. The program discovers regions of similarity amongst any set of DNA sequences and then draws a graphic that summarizes the length, location and relative orientations of any repeated sequences. Sequence similarity searching is a very general tool that forms the basis of many different biological sequence analyses but it is limited by the verbosity of traditional alignment presentation styles. Miropeats enhances the utility of conventional DNA sequence comparisons when looking at long lengths of sequence similarity by summarizing large-scale sequence similarities on a single page of PostScript graphics. Miropeats has been applied estensively to help understand shotgun assembly projects, to check cosmid overlaps and to perform inter-genomic comparisons.

A set of programs has been written to quantify the similarities between large numbers of cDNA sequences. This information is used to cluster similar sequences together. The main program can cluster thousands of cDNA sequences per day using a novel, computationally inexpensive algorithm. The clustering information is kept in a small index file so that disk storage requirements are negligible. Using this index file, subsidiary programs create various views and statistical summaries of the entire cDNA sequence collection.