Recurring Meeting of Cleveland Clinic - Stanford University

Date: November 22, 2013

Time: 4:00 PM EST

Means: Conference Call


  1. Ahmet Erdemir (Cleveland Clinic)
  2. Snehal Chokhandre (Cleveland Clinic)
  3. Craig Bennetts (Cleveland Clinic)
  4. Joy Ku (Stanford University)
  5. Henry Kwong (Stanford University)


  1. Discuss specifications and test problem development for the compute gateway to Simtk.
  2. Discuss potential high performance computing platforms (e.g XSEDE).
  3. Decide action items for next meeting.
  4. Other.

Immediate Action Items:

See notes for details.


  1. Discuss specifications and test problem development for the compute gateway to Simtk.

    • On the cloud computing prototype page, Craig described the files needed to run the simulations on the server. Depending on the options the user selects on the user interface, different sets of files will be used. File descriptions provide information about basic set of files, how to use the main script (which will run on the server) and other related files which are passed as argument for that script.
    • Craig also prepared a sample layout for web interface. This layout provides the basic idea of components of the web interface. Initially, only one available version of FEBio will be used. More versions may be made available in the future. There will be an option to select the model and depending on whether the model needs to be modified there will be two files associated with the model file, a script to modify that model and a configuration file that specifies what to modify in the model. Depending on whether the model was modified and run, a new model will be created which appends the _mod to the model file name and result files will be named similarly.
    • When the simulation is over, the user will be notified via e-mail. The notification e-mail should provide a link to download the results. An interface will be required for results retrieval.
    • All the files (except FEBio executable) will likely be in the subversion repository (all but the FEBio files). FEBio may be installed on the compute server (need to check potential licensing information). FEBio is free for academic use therefore users may need to agree before submitting jobs that their work if for academic purposes. The output files can be kept somewhere temporarily and do not need to be in the repository.
    • Additional post-processing scripts may be developed and provided on the cloud to extract relevant information from the result files and populate a database with that relevant information, which can be stored indefinitely.
    • An administration interface will also be developed for project administrators to add new models and related files. Location of example model files is provided in the description.
    • Models can be simplified for testing purposes. Memory / computing requirements will be evaluated for different cases as the development progresses.
    • Ahmet and Craig will provide the workflow used in the past while using Ohio Supercomputer Center or XSEDE.
  2. Discuss potential high performance computing platforms (e.g XSEDE).
    • A gateway can be created in XSEDE that will allow users to submit jobs using a web based interface, e.g., Robetta. Authentication and tracking requirements (of an XSEDE gateway) are already built in the operations. An application may need to be written to get support from XSEDE.
    • The prototype may be built on any machine for now. Google Compute Engine or Amazon Elastic Computing may also be used.
  3. Decide action items for next meeting.
    • Joy and Henry will work on adding upon the specification to include the administration panel and explore potential implementation of the interface.
    • Craig will provide a sample results interface (illustration).
  4. Other.
    • None noted.

RecurringMeetings/2013-11-22 (last edited 2016-05-04 22:09:49 by localhost)