New Features and Enhancements to Existing Features
A video overview of many of these features and enhancements can be found here.
New and Improved Modules & Pipeline Pane
The Modules & Pipelines pane has been redesigned to simplify searching and browsing. Improvements include:
A sliding panel to contain search/browse results
Support for "pinning" modules to a list of favorite modules (1) and maintaining a list of recently used modules (2).
To pin a module, simply drag it from the sliding panel to the area beneath Favorite Modules.
Access to a module's documentation from Modules & Pipelines pane (3).
Consolidation of Recent Jobs, Uploads and GenomeSpace tabs into the Modules & Pipelines pane (4).
Module Run Task Form Improvements
Support for running batch jobs over dragged in lists of files: users may now drag-and-drop files into an input file drop target (after checking the "Batch" box) to create lists of files over which to run a batch job. (Note that we've removed the Send Batch to <input-file-parameter> option from the Files tab directories. You may now simply drag and drop directories, after checking the Batch box, to add the files in the directory to the batch. You may also type in file paths, as you could before.)
Support for the parallel execution of jobs in a pipeline: for GenePattern Servers, like the Broad public server, that are configured to use a queueing system, independent jobs within a pipeline can run in parallel. This can decrease the time it takes a pipeline to run to completion.
Pipeline designer improvements to the linking of a module's output files to a downstream module's input file parameters. Prior to this release, one could only link a module's output file to follow-on modules' input file parameters by specifying the output file type or output file order. If your upstream module generated multiple files of the same type and their output order could vary between runs, you were unable to link unambiguously a particular output file to a downstream module's input file parameter. The pipeline designer now employs a combo box to establish links between a module's output files and downstream modules' input file parameters. The combo box lists all of a module's output file types (e.g., gct, cls) and output file order (1st Output, 2nd Output, ...); in addition, it permits the user to specify a regular expression that unambiguously identifies the desired output file.
The pipeline designer now supports attaching multiple category names to a pipeline. Thus, a pipeline can now appear under a functional category (e.g., RNA-Seq) as well as under the "pipeline" category.
When a pipeline exposes an optional module parameter as a prompt-when-run (PWR), the pipeline designer now gives the user an option of elevating that parameter to the required status.
Module Integrator Improvements
Users may add multiple support files at once to a module.
Users can create static file drop-down lists that reference module support files
Users can create dynamic file drop-down lists where list elements refer to directories rather than files. When selected, the entire contents of the directory will be uploaded. This new feature will benefit modules whose drop-downs require multiple files to be retrieved (e.g., BWA or Tophat genome indices).
Improvements in GenePattern Server Administration
SGE (Sun Grid Engine / Oracle Grid Engine) integration now works when GenePattern server is installed with a relative path to the jobs directory (the default installation configuration)
File purge intervals may now be set on a per-user and per-group basis, overriding the system-wide purge.
The 3.8.0 release fixes several reported issues in GenePattern, including:
The Administration menu was appearing in the Module Integrator's menu bar regardless of whether or not the user had admin privileges. This has been fixed in 3.8.0: the Module Integrator now only displays the Adminstration menu to users with adminstrator privileges.
A bug in GenePattern's SGE (Sun Grid Engine / Oracle Grid Engine) integration prevented job cancelation. SGE job cancellations now function correctly.
Send-to menus listed parameter names rather than prameter display names; they now list parameter display names.
The pipeline designer did not allow non-admin users with createPublicPipeline permission to create public pipelines; they now can.