Deploying and Maintaining Large Intermine Implementations: Phytomine at Phytozome

Phytozome (http://phytozome.jgi.doe.gov), the Joint Genome Institute’s plant genomics portal, enables users to interactively explore plant genomes, genes and gene families. While preconstructed bulk data files are provided, many users wish to assemble customized data sets based on specific filtering criteria, and including all (and only) data attributes of interest.

We provide that capability via PhytoMine, our local implementation of the InterMine data warehouse and web application (www.intermine.org).  PhytoMine, available at http://phytozome.jgi.doe.gov/phytomine, incorporates sequence, annotation, diversity (variation), expression, and gene family data for nearly 50 plant genomes. Users can transfer gene lists constructed in Phytozome to PhytoMine for additional filtering and download, or construct and manipulate lists directly.  Templates are provided to help users perform typical searches and data set downloads, while advanced users can construct their queries de novo using InterMine’s query builder interface.

While InterMine is used by many primary genome data providers around the world, JGI’s implementation is relatively unique as a public mine in terms of the number of complete genome data sets hosted (48), as well as the density of some of the included data sets (e.g., a poplar diversity data set with more than 500 individuals).  We discuss some of issues that arise with very large, frequently updated datasets, when seeking to provide a highly available, correct, and reasonably responsive web-based data service, in terms of schema design, query optimization and loading protocols. We also provide an overview of the data types and advanced search, display and download capabilities available at PhytoMine.