Appion

r15725 contains what I have thus far:

apRefineJob.py is the base class for the Job class we talked about.

apRefineJobEman.py is a working subclass of it

apRefineJobFrealign.py does every tasks it needed for the iterations but not the tasks needed before and after the iterations.

runRefine.py is the one that should be a subclass of agent class which need to parse jobtype so that the choices of the refinejob can be made. Right now, the selection is hard coded in it.

If self.jobtype is set in runRefine.py to emanrecon, the following command will produce an output equivalent to emanJobGen.php default from the example preparation I made for #1342 and likely to be runable on guppy (not tried).

runRefine.py --safedir=/ami/data00/appion/zz07jul25b/recon/eman_recon1/ --recondir=recon --stackfile=start.hed --modelfile=model.mrc --totalpart=7972 --boxsize=80 --apix=3.26 --startiter=1 --enditer=1 --sym=d7 --hard=25 --classkeep=0.8 --classiter=8 --jobtype=emanrecon --setuponly --appionwrap=/opt/bin/appion

If self.jobtype is set in runRefine.py to frealignrecon, the following command will produce shell scripts like prepFrealign,py. This is not runable.

runRefine.py --recondir=. --stackfile=start.hed --modelfile=model.mrc --totalpart=500 --boxsize=80 --apix=1.6 --startiter=2 --enditer=4 --sym=c1 --setuponly

r15764 is a non-working apRefineJobXmipp.py I think several global function such as locateXmippProtocol should go into appionlib so that it may be reusable for ML3D case.

The xmipp refine is running, so I'm going to assume this is done.