GenePattern 3.3.1 Release Notes

Released March 1, 2011

GenePattern 3.3 contains several new features and enhancements, as well as bug fixes and beta features.

New Features

Module Search

A search box on the home page allows users to find modules by typing names or parts of names. The results are presented in a dynamic Google search style.

Uploaded Datasets

Users can now view all files they have uploaded and use them in any future analysis via the Uploads tab on the home page.

Job Output Subfolders

Users can now browse files in subfolders on the Job Status page, enabling users to directly use those files as module input. In addition, users can download jobs with the subdirectories preserved.

Fixes

The 3.3 release fixes issues in several functional areas of GenePattern, including:

 

  • Pipeline execution
  • Job execution
  • LSF integration
  • GenePattern server configuration
  • Other fixes

Pipeline Execution

GenePattern 3.3 extensively refactors the GenePattern pipeline engine to fix the following issues:

 

  • pipelines handle optional input parameters without errors
  • output files from nested pipelines are passed to the outer pipeline
  • pipelines are checked for missing modules before they run, eliminating this as a source of errors
  • pipelines run asynchronously, allowing pipeline jobs to resume after a server restart
  • running pipelines can be terminated without leaving child jobs in invalid states, reducing resource use
  • pipeline status is shown correctly on the Job Status page for terminated or failed pipeline runs
  • failed jobs within pipelines are shown as errors
  • execution of pipelines is streamlined to reduce resource consumption

Job Execution

GenePattern 3.3 provides improvements to job execution and configuration:

 

  • job configuration file errors are displayed on the server settings job configuration page
  • job analysis is prevented from using all the available database connections

LSF Integration

GenePattern 3.3 improves the integration between GenePattern and the LSF batch queuing system:

 

  • jobs are not permanently stuck in invalid states if the GenePattern server encounters problems submitting jobs to LSF
  • the error messages are more informative when errors occur with LSF job submissions
  • LSF jobs can be terminated immediately, rather than being queued for termination
  • a pending LSF job can be terminated

Server Configuration

GenePattern server configuration is improved in this release, allowing server administrators to:

 

  • reload the job configuration file programmatically without a server restart
  • configure properties based on users and groups
  • suspend or start the job queue from the job configuration page
  • customize the location of the jobs directory
  • customize the location of each user's uploads directory
  • prevent new jobs from being added if disk space is almost full

Other Fixes

GenePattern 3.3 includes a number of other bug fixes and improvements:

 

  • the GenePattern web front end can better handle running under a heavy load, providing shorter load times for the Pipeline Designer and Job Results pages
  • modules do not fail when the GenePattern server is installed in the default location on a 64-bit Windows machine (C:Program Files (x86)GenePatternServer) (Note: See Known Issues for the exception.)
  • the GenePattern server retains custom server variable values on restart
  • the Pipeline Designer page sorts modules versions numerically, rather than alphabetically
  • more than one job can be deleted from the Job Summary page at a time
  • request parameters are preserved even if there is a login dialog as an intermediate step
  • modules retain input file specifications on reload even if the parameter names are similar
  • pop-up menu contents for the job results files present correct options
  • the GenePattern installer can handle port specifications

Beta Features

These features are in development. While they are present in the GenePattern 3.3 code, they are turned off by default. If you would like to help us test them, please contact us at gp-help@broadinstitute.org for instructions for turning these features on in your GenePattern server installation.

Advanced Upload for Large Input Files

This feature implements a new file upload system for handling input files larger than 2 GB. This is available in individual modules (select the Advanced Upload radio button) and the Uploads tab, and enables batch execution and the directory input parameter, as described below.

Batch Execution

Users can select multiple files in a folder via the Advanced Upload feature. The module will be run once for each file selected.

New Input Parameter: directory

Module developers can specify an input parameter as a directory. The user can then select a directory of input files for the module, rather than a single input file. All the files in this directory are treated as input for a single run of the module. Using this parameter type invokes the Advanced Upload option.

Upload Files via the Uploads Tab

Users can upload files without running a module by using the Upload Files button in the Uploads tab. Uploaded files can used in any future analysis.

Important Note: When users run a job normally, the job input files are scheduled to be purged 7 days after the job run (on the Broad public server and on servers where the default time is in place). However, when users upload a file directly, the files are not scheduled for deletion at all. Uploaded files must be manually deleted from the server.

Known Issues

Known issues in GenePattern 3.3 include:

  • ComparativeMarkerSelection will fail when run on a GenePattern server installed in the default location of a 64-bit Windows machine. Other Java modules that invoke R with the command line " -DR= ..." are likely to fail as well. Workaround: Edit the R 2.5 Home setting on the Programming Languages page to change to the default GenePattern install location (C:Program Files (x86)GenePatternServer).
  • GenePattern does not support module names with spaces in them.
  • Some GenePattern modules do not work when the input filename contains a space character. Workaround: Rename the input file without a space character.
  • On Macintosh: Java Runtime Environment (JRE) 1.6.0_15 does not work. Workaround: The GenePattern team suggests updating to the most recent JRE (GenePattern 3.3 has been tested with 1.6.0_20 and above)
  • On Windows 7/Vista: The StartGenePatternServer and StopGenePatternServer applications must be run as an administrator. To start or stop the GenePattern server, right-click on StartGenePatternServer.exe or StopGenePattnerServer.exe and select Run as administrator.
  • For GenePattern administrators who have changed the "input.file.mode" property (in the genepattern.properties file) from default: GenePattern 3.3 only works with the default setting, input.file.mode=path.