sxlocres - Local Resolution: Compute local resolution of a map.
Usage in command lines:
sxlocres.py firstvolume secondvolume maskfile outputfile --wn=window_size --step=shell_step_size --cutoff=res_cutoff --res_overall=overall_resolution --radius=radius --fsc=output_file_name --MPI
sxlocres exists in for both MPI and non-MPI versions.
mpirun -np 32 sxlocres.py refinementdir/main002/vol0.hdf refinementdir/main002/vol2.hdf locres.hdf --cutoff=0.5 --radius=28 --res_overall=0.3 --fsc=fscvol.txt --MPI
- First half-volume: (default required string)
- Second half-volume: (default required string)
- 3D mask: Defines the region where the local filtering should be applied. (default none)
- FSC window size: Defines the size of window where the local real-space FSC is computed. (default 7)
- Fourier shell step size [Pixels]: (default 1.0)
- Resolution cut-off [1/Pixel]: Cut-off for the FSC. By default it goes to the Nyquist frequencies. (default 0.5)
- Mask radius [Pixels]: In case no mask is provided, a hard sphere of this radius will be used. (default -1)
- Overall resolution calibration (0 to 0.5). See Description below for details. (default -1)
Output volume: Each voxel contains the resolution for this area in absolute frequency units. (default required string)
- FSC output file: Contains the overall FSC curve computed by rotational averaging of local resolution values. It is truncated to res_overall. (default no curve)
- Local resolution values are computed within small windowed areas (size wn^3). Due to small sample size the values are inaccurate and tend to underestimate resolution. Therefore, option res_overall allows to calibrate the average value of all local resolutions to the overall resolution, as estimated by the FSC between two full size volumes. I This value has to be independently obtained by the user - for example using sximgstat.py to compute (masked) FSC from half-volumes and reading the absolute frequency crresponding to 0.5 (or other) cut-off value.
Author / Maintainer
- category 1
- works for most people, has been tested; test cases/examples available.
None. It is perfect.