geWorkbench

Using geWorkbench

Q. Can I run geWorkbench on a 64-bit Windows system?

A. Yes. As of release 2.0.0, geWorkbench has been tested on both 64-bit Windows 7/Vista operating systems, as well as 32-bit Windows XP. The formal release testing for Windows is done using the 32-bit Java JRE6. However, geWorkbench has also been tested with the 64-bit Java JRE6 on the 64-bit Windows 7 platforms, and only minor compatibility issues have been found. These are listed in the Known Issues section.

The prebuilt installers with JRE6 all include the 32-bit JRE. If you wish to try the 64-bit JRE, please use a version of the geWorkbench installer which does not include the JRE. Further instructions can be found on the Download and Installation page.

Q. Will calculations run faster using a 64-bit JVM than on a 32-bit JVM?

We have compared single runs of ARACNe on 32-bit and 64-bit JVMs and found a significant speed increase when using the 64-bit JVM.

Testing conditions:

• Dataset: We tested using a dataset with 176 microarrays of type HG-U133A, with 22,283 probesets. A set of 2013 hub markers was used.
• ARACNe parameters: p-value cutoff 0.01, Bonferroni correction, DPI 0.15, no bootstrapping.
• Machines: Core 2-6700 CPU with 4 or 6 GB memory, 2.66 GHz.
• OS: Windows 7, 64 bit Enterprise Edition.
• JVM: Oracle 1.6.0_32 or 1.6.0_31.

Results: On both machines tested, the ARACNe jobs always finished faster when running on the 64-bit JVM than on the 32-bit JVM. In paired, back-to-back tests on each machine, the time to finish was 20% to 90% longer on the 32-bit JVM (Actual results from 3 tests: 21%, 51%, 95% longer on 32 bit).

Possible explanation: Current AMD and Intel CPUs, when operating using the 32-bit x86 instruction set, have access to 8 general purpose registers. However, when operating in 64-bit mode (x86-64), the operating system has access to an additional 8 general purpose registers (1). These may provide an advantage to code with tight loops in the calculation (2). For 64-bit operation, both a 64-bit operating system and a 64-bit JVM must be used.

References:

1. http://en.wikipedia.org/wiki/X86-64#Architectural_features

2. http://en.wikipedia.org/wiki/64-bit#Pros_and_cons

Q. I get a Java error when I try to start geWorkbench.

A. This is almost always caused by the Sun Java JRE not being installed or not being found (when you use a version of geWorkbench that does not include the JRE). You can either try a geWorkbench installer that includes the JRE, or make sure that an appropriate JRE is installed on your system.

Q. Where can I obtain the latest annotation files for my microarray platform?

A. Affymetrix annotation files can be downloaded from their support site, at www.affymetrix.com. A free Netaffx account sign-up is required.

There are several ways to reach the annotation files, two of which are shown next:

1. The following link will take you directly to a list of most (but not all) current annotation files.

http://www.affymetrix.com/support/technical/annotationfilesmain.affx

2. Some arrays, for example the HG-U95Av2 array used in some geWorkbench tutorials (for the BCell-100.exp data set), are not shown in the above list. However, the files can be obtained from the catalog of individual array products at the following link:

http://www.affymetrix.com/support/technical/byproduct.affx?cat=arrays

For the HG-U95Av2 annotation file, browse down to 3' Gene Expression Analysis and look for Human Genome Arrays. Then find the section "Current NetAffx Annotation Files".

The file format required is CSV (comma separated values).

Q. Is the Affymetrix Gene 2.0 ST annotation file type supported?

A. Yes. geWorkbench 2.4.0 added a parser to support the Affymetrix Gene and Exon 1.0 ST transcript-level, CSV-format annotation files. The Gene 2.0 ST transcript-level CSV annotation file uses the same format and can be read in using the Gene/Exon 1.0 ST parser choice in geWorkbench.

Q. How do I increase the amount of memory available to Java to run geWorkbench?

A. It depends on how you are running geWorkbench.

1. If you are running a packaged distribution of geWorkbench (created using InstallAnywhere), there is a file in the geWorkbench root directory called UILauncher.lax. There is a line there which specifies the Java heap size:

lax.nl.java.option.java.heap.size.max=1073741824

Here it is shown set to 1 GB. You can experiment with increasing this, subject to the amount of memory in your machine and demands on it from other applications.

2. If you are running geWorkbench from the Generic or source distributions using Ant, you can edit the build.xml file found in the geWorkbench root directory to alter the memory requested using the variable jvmarg:

<target name="run" depends="init" description="Runs geWorkbench.">
    <java fork="true" classname="org.geworkbench.engine.config.UILauncher">
       <jvmarg value="-Xmx1024M"/>
       <jvmarg value="-Djava.library.path=lib"/>
       <arg value="all_release.xml"/>
       <classpath refid="run.classpath"/>
    </java>
</target>

Here it is shown requesting 1 GB.

Q. Is there a shortcuts menu for available commands?

A. Yes! - on the PC, you can use F12.

Q. geWorkbench appears to be frozen.

A. Sometimes, a "modal" dialog box can appear but by chance get hidden behind other windows (if the user clicks on some other window without noticing the dialog box). The hidden dialog box is waiting for user interaction, and geWorkbench is waiting for the dialog to be dismissed, so the application is unresponsive. This is a feature of how Java works. One way to see if this has happened is to minimize all open windows on your desktop, for all applications, and the maximize the geWorkbench window. If there is an open dialog box, it should now appear in front of the geWorkbench main window. (c.f. Mantis entry 1959).

Q. How can I download the geWorkbench source?

A. The latest geWorkbench source code can be downloaded from Github. Please see the instructions on the Download the geWorkbench Source Code section of the Download and Installation page. Instructions for compiling the code are available on the same page at Compiling the geWorkbench Source Code.

Q. How do I turn on logging in installer-based versions of geWorkbench?

Copies of geWorkbench installed using the InstallAnywhere-generated install file contain a configuration file in their root directory called UILauncher.lax. Make the changes like the following to turn on logging to files:

#   LAX.STDERR.REDIRECT
#   -------------------
#   leave blank for no output, "console" to send to a console window,
#   and any path to a file to save to the file

lax.stderr.redirect=log/stderr.log

#   LAX.STDOUT.REDIRECT
#   -------------------
#   leave blank for no output, "console" to send to a console window,
#   and any path to a file to save to the file

lax.stdout.redirect=log/stdout.log

Q. How can I reference geWorkbench in a publication?

A1: geWorkbench is described in the following publication:

http://www.ncbi.nlm.nih.gov/pubmed/20511363

Floratos A, Smith K, Ji Z, Watkinson J, Califano A. (2010). geWorkbench: an open source platform for integrative genomics. Bioinformatics 26(14):1779-80. Epub 2010 May 28.

A2: Further support statement:

"Analysis was carried out using geWorkbench (http://www.geworkbench.org), a free open source genomic analysis platform developed at Columbia University with funding from the NIH Roadmap Initiative (1U54CA121852-01A1) and the National Cancer Institute".

Q. What differences are there in how genes without symbols are handled in GO Terms analysis vs a CNKB query?

A. In the GO Terms analysis component, only genes with a gene symbol are used in the analysis, and only genes with a gene symbol will appear in the list of genes associated with a selected term. By contrast, in the CNKB component, markers which match the particular id (e.g. Entrez ID) used in the query are returned even if they do not have an associated gene symbol. (Mantis issue #2478).