New (2021) SPT refinement protocol including subtomogram, subtilt and defocus refinement. For a simple run, use
	e2spt_refine_new.py --ptcls sets/ptcls.lst --ref reference.hdf --iters p,p,t --startres 50 --goldstandard
	Control refinement modes with the --iters option. See details in the help below. 
	The program needs a resolution to be specified with --startres, which will be the maximum resolution considered in the first iteration. In later iterations, the maximum resolution is calculated from the FSC of the previous iteration. When --goldstandard is specified, the reference is phase randomized to the --startres resolution. 
	One major difference of the new protocol is that the program now can model the localized 2D particle motion by considering the motion trajectory of each particle along with its neighbors. For each particle, --smoothN controls how many of its neighbors are considered to model the local motion, and --smooth controls how much the neighboring particles are weighted during the alignment. The weight of neighboring particles decays in a Gaussian form based on the distance to the center particle of consideration. --smooth=0 means only the center particle is considered, and the program should perform in a similar way as the original subtilt refinement.

Option Type Description
--version None show program's version number and exit
--ptcls str 3d particle input
--ref str reference map
--startres float Starting resolution for the refinement in A. Default is 50. This will be the maximum resolution considered for the first iteration. In later iterations, the maximum resolution is calculated from the FSC of the previous iteration (unless --maxres is specified).
--goldstandard None Phase randomize the reference to the starting resolution (--startres) independently for the even/odd subsets of particles.
--goldcontinue None Continue from previous gold standard refinement. Ues the _even/_odd version of the given reference.
--path str Directory of the refinement.
--sym str symmetry
--iters str Iteration information. Input types of refinement separated by comma. p - 3d particle translation-rotation. t - subtilt translation. r - subtilt translation-rotation. d - subtilt defocus. Default is p,p,p,t,p,p,t,r,d. Character followed by number is also acceptable. p3 = p,p,p
--keep str Fraction of particles to keep. Note this is controlled at three separate steps. When default --keep=.95, it removes the worst 0.05 3D particles, 0.05 2D subtilt with the worst score, and 0.05 of subtilt with the largest drift. Also accept comma separated values (0.9,0.5,0.5) to set different keep thresholds for the three classes
--setsf str structure factor for sharpening
--tophat str Options for filtering maps. Run 'e2help.py tophat' for more information. Default=wiener.
--ssnrwt None weight particles during reconstruction by SSNR accroding to references.
--curve None Filament refinement mode. still under testing
--use3d None Use projection of 3d particles instead of 2d sub tilt series. This may be more useful for thicker sample but can be significantly slower.
--localrefine None only perform local search around the solution from the last iteration
--loadali2d str load previous 2d alignment from an aliptcls2d_xx.lst file
--loadali3d str load previous 3d alignment from an aliptcls3d_xx.lst file
--maxres float Maximum resolution to consider in alignment (in A, not 1/A). The program will determine maximum resolution each round from the FSC of the previous round by default.
--minres float Minimum resolution to consider in alignment (in A, not 1/A)
--mask str Mask applied to the results (instead of automasking)
--automaskexpand int Default=boxsize/20. Specify number of voxels to expand mask before soft edge.
--preprocess str Preprocess each 2-D subtilt while loading (alignment only)
--parallel, -P str Run in parallel, specify type:<option>=<value>:<option>=<value>. See http://blake.bcm.edu/emanwiki/EMAN2/Parallel
--threads int threads for post-processing
--breaksym str Specify a symmetry to break
--maskalign str Mask file applied to 3D alignment reference in each iteration. Not applied to the average, which will follow normal masking routine.
--maxshift int maximum shift. default box size/6
--maxang int maximum angle difference from starting point for localrefine.
--smooth float smooth local motion by this factor. smoother local motion with larger numbers. default 100
--smoothN int number of neighboring particles used for smoothing. default 15
--m3dthread None do make3d in threading mode with shared memory. safer for large boxes
--maxtilt float Excluding tilt images beyond the angle
--ppid int Set the PID of the parent process, used for cross platform PPID
--verbose, -v int verbose level [0-9], higher number means higher level of verboseness