[phenixbb] Composite Omit Map with Parallel Processing

Nathaniel Echols nechols at lbl.gov
Tue Aug 19 12:28:11 PDT 2014

On Tue, Aug 19, 2014 at 10:15 AM, Whitley, Matthew J <mjw100 at pitt.edu>

>  I am trying to calculate a simulated annealing composite omit map via
> parallel processing in Phenix 1.9-1692.
> Our computer cluster uses a PBS que, and jobs are submitted using the
> 'qsub' command.
> In the 'Processes' tab under Phenix Preferences, the queueing system is
> set to PBS, and the queue submit command is set to qsub + additional flags
> (walltime, etc). All the usual boxes are checked to indicate to Phenix that
> jobs should be run in parallel when possible.
> However, when I actually run the composite omit map calculation, the
> specified qsub flags do not get carried over into the calculation
> parameters. If I click on Settings-->View all settings in the Composite
> Omit map calculation window, the qsub_command parameter under 'parallel' is
> set to None.
> Because my desired walltime is not being carried over with the qsub
> command when the calculation starts, the calculation for each individual
> box is aborting prematurely when it reaches the default process time limit
> set by our cluster administrator.
> In the composite omit map calculation in previous versions of Phenix, when
> clicking "Run" you used to be able to then click "edit parameters and run"
> in order to check all settings one last time before starting the job. This
> option does not seem to be available in Phenix 1.9-1692, and thus I can't
> verify and/or edit the qsub command directly before firing off the job.

FYI, this is a completely new program - no relation to the old one (aside
from sharing some underlying code, i.e. phenix.refine).

> Any ideas on how to overcome this? Am I missing something?

Yes, but it's mostly my fault for not making this more obvious and also not
propagating the preferences setting to the program option.  If you go the
Settings menu and select "Search parameters" and enter "qsub", it will pop
up the control for the qsub_command parameter.  But I have also modified
the GUI so it will behave as you expected in the next build.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20140819/78b3d245/attachment.htm>

More information about the phenixbb mailing list