Genome Annotation in the Cloud through XGDBvm Virtual Server Instances Deployed at iPlant

Annotation is often a rate-limiting step for newly-assembled eukaryotic genomes, requiring complex workflows, specialized software and access to high performance computing (HPC) facilities. We sought to bridge the gap between assembled data and a fully annotated, browsable genome by building xGDBvm, a feature-rich genome annotation pipeline and browsing platform on a Linux-based virtual server. We report the initial deployment of xGDBvm as a virtual machine (VM) in iPlant's Atmosphere Cloud Service, with programmatic links to HPC for added processing power.  An xGDBvm-iPlant instance is configurable in a few simple steps and can be accessed over a secure Web connection at a unique IP address. xGDBvm accesses the user's input data via the iPlant Data Store cloud storage service, and it can send computationally-intensive processes to remote HPC facilities at Texas Advanced Computing Center (TACC) using iPlant's Discovery Environment Web services. xGDBvm masks genome sequences, computes spliced alignments from transcript and/or protein inputs, predicts gene structures and gene quality scores, and indexes all sequence data for BLAST. xGDBvm ultimately packages output data on a user-controlled virtual storage volume and displays output as a full-featured genome browser that can be either public or password-protected on the Web. A variety of search, download and sequence analysis tools are built in to xGDBvm’s genome browser, and contextual help is provided. Questionable gene predictions can be re-annotated manually using the integrated yrGATE Community Annotation tool. xGDBvm can also be configured to update (append or replace) existing data, or to load pre-computed spliced alignments or annotations. Multiple genomes can be annotated in a single xGDBvm instance, and output can be archived and restored. Any registered iPlant user can launch an xGDBvm instance at http://atmo.iplantcollaborative.org, by searching under the name 'xGDBvm-beta'. An xGDBvm project page, wiki, and hosted genome dataset can be found at http://goblinx.soic.indiana.edu.