Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

BEAST analyses done through the DE are run on the Stampede supercomputer at the Texas Advanced Computing Center, which has queue. If there are many jobs in the backlog, which can happen randomly, but especially after they have had a service outage, there may be a long wait time in the queue. This app is set up to run for the maximum run time (48 hours) so that it can handle fairly large analyses. Because of the long run time, it does not move as quickly through the queue. Therefore, for very small jobs, when you combine the wait time and the processing time, it may not be much faster to run BEAST on through the DE than on a desktop computer, but for large jobs, it should be much faster.

Quick Start

  • To use BEAST 2.1.3, upload an XML file generated by BEAUti 2.1.3.
  • See the BEAST help page for more information on using BEAUti.
  • Click on the app "BEAST" under the High-Performance Computing category.
  • Rename your analysis something meaningful. You can add comments and change where the results are stored if you want. BEAST will by default save your input file, so it doesn't matter if you check "save inputs".
  • Click on the input pane and select your XML input file. Note that XML files generated by older versions of BEAUti will not work.
  • If you want to change or add parameters, click on the parameters panel and add each one separated by a space. Be sure to format the parameters as specified for running BEAST on the command line (see Resources, below).
    • This is a High Performance Computing (HPC) application designed to run on a graphical processing unit (GPU) on the Stampede supercomputer at the Texas Advance Computing Center. Because of this, it will generally run the fastest using the default parameter (-beagle_gpu).
  • Click the Launch Analysis button.
  • There is a 24 hour time limit on this application. For large jobs, it is recommended that you first submit and XML file with a small number of generations (e.g., 1 million) to determine how many generations your analysis can get through in 24 hours and then set up your main analysis accordingly. Jobs that do not completely finish should still give output at the end of 24 hours.

...