The Aga (Automated genome analysis) system is a custom implementation of Microbase v1.0 to form an automated genome comparison pipeline, together with service-based interfaces to computationally expose the resulting all-against-all dataset and a Web portal for interaction by a biologist. A number of custom responder components have been developed, that populate an object-oriented data warehouse, which can be queried and viewed using genome comparison browsers.

Aga not only serves as an example for the bioinformatician interested in deploying and customising a Microbase instance, but also provides a useful tool for a biologist. An overview of the Aga architecture is shown below

Complete genome sequences are sourced as they appear, from the genome reviews ftp site at EBI, parsed and cached, and processed by a number comparison algorithms wrapped as responders. A listener component (actually composed of two responders) scans the remote site and generates a ‘new genome event’ when a new genome is detected. A genome is fetched from the site in response to this event and passed to a custom responder termed the ‘Genome Pool’ that fetches the EMBL record and parses it into a BioSQL database. The Genome Pool is a standalone component with Web service interfaces to allow genomes and their features to be computationally accessed by a remote client. As such it forms a useful service enabled genomic databases for a range of Bioinformatics Grid applications outside of Microbase.

A series of ‘System’ responders carry out all-against-all comparisons on all genome sequences, fetched from the Genome Pool, at the nucleotide and protein sequence levels. These algorithms include BLASTp, BLASTn, mummer, Promer and Ssearch. The algorithms are run as jobs that are farmed across a computational grid. In this case the available Grid varied in size from the 96 CPU’s of a dedicated cluster through to the Newcastle campus grid, comprising up to 2,000 machines. The large datasets that result are cached as torrent files that remain distributed across the torrent servers in the network, until they are retrieved and parsed by further responders, that populate an object-oriented data warehouse, when network traffic allows. The comparison data in the warehouse may be accessed computationally via a Web service interface, or manually by a user, through a set of visualisation and query tools in a Web portal. User responders capture user queries to be performed over the warehouse (such as automatic searches for gene homologues) and automatically notify a user when results are generated.

Please visit the Microbase Lite genome analysis site (temporarily unavailable due to server reconfiguration)