VectorBase uses the Genome Model Organism Database (GMOD) construction set for the storage of genome sequence and annotations. The GMOD CHADO schema facilitates the rapid incorporation of diverse data types, e.g. literature, controlled vocabularies and good inter-operability with the manual annotation effort, using Apollo, as well as the Ensembl system. For web display and access to the genome data and annotation, VectorBase utilises the Ensembl database schema, API and web code (⁴).

The Ensembl system provides a good model for handling genomic data from a number of species in a consistent and unified manner and a highly sophisticated set of interlinked web pages.

Entry points into the genome are through text searches of gene names, symbols and descriptions, pairwise similarity searches and from cross-references in the public nucleotide and protein databanks.

The VectorBase website contains standard Ensembl style gene and transcript pages. Gene pages contain information about the prediction including gene orthologue and protein features (signal peptides, trans-membrane domains, InterPro domains). Gene Ontology (GO) codes and Enzyme Classification (EC) numbers are assigned where possible.

Batch file downloads are available for both the raw sequence data (fasta files of genomic sequence, including repeat masked sequence and ESTs) and the annotation (GFF3 files or a MySQL dump for use with the Ensembl API).

Batch searching capabilities are handled by the powerful data mining tool BioMart (⁵) and through two spreadsheets, AnoXcel and AegyXcel (⁶) that contain gene based information about the presence of signal peptides, trans-membrane domains, protein domains and the best similarity with yeast, Drosophila and human.

The two genomes currently available through VectorBase are good examples of the multiple roles undertaken by the group.

The A.gambiae PEST genome was published in 2002 (⁷) with a genome size of 260 Mb and ∼14 000 genes. The predicted gene set has been reviewed and updated several times since publication. This process involves a blend of automated evidence-based gene prediction (⁸) and manual approaches. Manual appraisal of gene models is firstly targeted to regions of interest from the community and regions for which we are aware that automated approaches fail. A manual re-annotation of chromosome arm 2L is finished. Manually appraised gene models are highlighted for the user as a separate track on the genome browser.

The A.aegypti Liverpool strain was sequenced and assembled by The Institute of Genomic Research (TIGR) and the Broad Institute. Aedes has a large genome size of 1.3 Gb and a predicted gene complement of ∼16 000 genes. This represents the first-pass annotation of the genome using automated approaches. Improvements in quality will be achieved by manual efforts and enhancements to the evidence-based automated gene predictions.

The presence of two related mosquito genomes in VectorBase allows for comparative analysis to identify conserved regions between the genomes. This can be useful in verifying, and correcting, gene models and for studying gene family expansions. Translated BLAT (⁹) similarities between the two mosquito species and with Drosophila have been identified. Figure 1 shows a view of an orthologous locus between the two mosquito genomes and highlights the expanded intron size in A.aegypti and related higher repeat content.

The DAS protocol (¹⁰) allows community researchers to integrate and display their data sets in the genome browser window. This is especially powerful with alternative sets of gene predictions. As an example, the Anopheles browser contains DAS tracks for alternate EST based gene predictions from AnoEST (¹¹), an independent re-annotation effort by Li et al. (¹²) and predictions based on mass spectrometry data (¹³).

Microarray data exists for both Anopheles and Aedes and array probes from both species are mapped to the genome. These alignments are displayed in the browser and queries can be made against these via BioMart. Concise expression summaries for probes and genes are made available as experiments are published.