Separate Into Classes : Separates stacks of particle images into stacks for each class and evaluates results.

Usage in command line (beta version):

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --align_isac_dir=isac_or_beautifier_dir --filtrad=filter_radius --apix=pixel_size --shrink=shrink_factor --ctf=ctf_mode --nvec=number_of_eigenimages --pca_radius=radius --chains_radius=radius --chains_exe=spchains_executable --applyparams=centering_mode --write_centered --debug --imgformat=image_format --verbosity=verbosity_level

Regular ISAC (top) vs. Evil ISAC (bottom)

The purpose of sp_separate_class.py is to:

: write particle-membership lists for each class
: write separate stacks for each class, 
: optionally low-pass filter and/or downsample the images, and
: optionally compute eigenimages (basis images) for each class

1. Standard usage: create separate stacks for each class:

sp_eval_isac.py input_class_avgs output_directory 

2. Apply a low-pass filter to the image stacks:

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --filt=filter_radius --apix=pixel_size

Filter radius is in units of Angstroms. If –apix is not specified, program will assume units of pixels^-1.

3. Downsample output image stack:

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --shrink=shrink_factor

4. Apply ISAC alignments to particles:

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --align_isac_dir=isac_or_beautifier_directory

Alignments used by ISAC or beautification will be applied to the particles. In addition, the average and variance for each map will be written.

5. Compute eigenimages (basis images) for each class:

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --align_isac_dir=isac_directory --nvec=number_of_eigenimages --pca_radius=radius

In addition to the average and variance, the requested number of eigenimages will be computed also. If –pca_radius is not provided, the whole image will be used to compute the eigenimages.

6. Apply centering to each class as determined by sp_center_2d3d.py:

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --align_isac_dir=isac_directory --write_centered --applyparams=centering_mode

If you ran sp_center_2d3d.py, you can also apply those centering parameters to the individual particles. The –write_centered flag will write out the particles; omitting the flag will simply write out the alignment parameters without applying them. The options for –applyparams are 'intshifts' for integer shifts and no rotation (i.e., no interpolation) and 'combined' (rotation and shifts).

7. Apply a Gaussian band-pass filter to an image stack (e.g., class averages):

Shape of the band-pass filter

sp_eval_isac.py input_class_avgs output_directory --particles=input_image_stack --align_isac_dir=isac_directory --write_centered --applyparams=centering_mode

The default beautifier settings (CTF-correction using a Wiener filter and power-spectrum adjustment) will amplify the low-resolution data, and may make the averages more difficult to interpret. A band-pass filter will dampen the lowest-resolution data, and also the high-resolution noise. This filtration may be helpful for recognizing smaller complexes.

(top) Beautified images and (bottom) band-pass filtered images, centered at 14 Angstroms

input_class_avgs

Set of 2D class averages, with particle-membership information in header. (default required string)

Filtered, aligned particles of TcdA1 corresponding to class #9 below

output_directory

Directory where outputs will be written. (default required string)

--particles

Particle image stack. (default required string)

--filtrad

Gaussian low-pass filter radius, Angstroms if apix specified below, else pixels^-1. (default None)

--apix

Angstroms per pixel, might be downsampled already by ISAC2. (default None)

--shrink

Downsampling factor, e.g., 6 → 1/6 original size. (default None)

--align_isac_dir

If applying alignments, directory for ISAC output (default None)

--nvec

Number of eigenimages to compute. (default None)

--pca_radius

Radius (pixels) to use for computation of eigenimages. (default None)

--ctf

Applies CTF correction: 'flip' for phase-flipping, 'wiener' for Wiener filter. (default None)

--bandpass_radius

Radius for the center of the band-pass filter, Angstroms.

--bandpass_width

Width of the band-pass filter, units absolute frequency (i.e., px^-1).

--verbosity

Controls how much information to write to screen. (default 1)

--chains_radius

Runs sp_chains.py internally to order input class averages. (default None)

--chains_exe

Path for sp_chains.py if not the default. (default None)

--imgformat

Format of optional output aligned-imaged stacks. (default .mrcs)

--debug

Tests –applyparams=intshifts by applying rotation and integer shifts separately and computing average. (default False)

classmap.txt

Class-to-particle lookup table, one file for all classes

For each class (from left to right): average, variance, and first 5 eigenimages

docclass???.txt

List of particles for each class, one file per class

bdb:stkclass_???

Virtual stacks of particles for each class

bdb:stkflt_???

(Optional) Virtual stacks of filtered particles for each class

params_???.txt

(Optional) Particle alignment parameters, one file per class

stkavgvar.hdf

(Optional) Montage of class averages and variances

stkeigen.hdf

(Optional) Montage of class averages, variances, and eigenimages

: Should allow filter types other than Gaussian low-pass

Tapu Shaikh

Category 1:: APPLICATIONS

sphire/bin/sp_eval_isac.py

Beta:: Under evaluation and testing. Please let us know if there are any bugs.

There are no known bugs so far.