Sequence of steps necessary to detemine single particle structure using cryo data (with CTF estimation)

Step 1

  1. Convert scanned tiff micrographs into hdf files. The files can be examined using

    • i. Run .

    • ii. Set foc=f if micrographs are scanned from films.
    • iii. At this stage, one has to know the pixel size `"""Pixel_size"""`  of scanned micrograph, or CCD frames.

    • `"""Pixel_size"""=("""Scan_step_size""")/("""Magnification""")`

    • where Scan_step_size is the scan step size of the scanner, or physical pixel size of CCD camera (Scan_step_size of CCD in Polara is `15*10^(-6)m`). Nicon scanners have pixel size : `6.35*10^(-6) m`.

  2. Calculate power spectra of all micrographs, examine them, delete micrographs that show drift, have no Thon rings and so on...
    • i. Calculate power spectrum using Welch's periodogram: Run .

    • ii.Properly set mask radius for exclusion of the central peak in the power spectrum (10 pixels or more ) will help to identify drift or astigmatism in the power spectra.
    • iii. 1-D rotationally averaged power spectra (have roo_ as prefix) can be used for visualization.
  3. Calculate defocus of each micrograph from 1D rotationally averaged power spectra, and store the defocus value in the header of micrographs by running

  4. Run the automated particle program, particles from each micrograph are in a separate stack file in spider format.

    • i. Run

    • ii.The template file has to be created before using this command.
    • iii. Properly setting n_plt helps to reduce the number of bad particles.
    • iv. Select a noise image from an arbitary micrograph. Save the coordinate file in text tile as


  1. Screen all stack files and delete windows that do not contain particles. This is done using program v2:

    • i. v2 <stackfile>

    • ii. middle-click on the image display
    • iii. select 'del' mode from the instpector window
    • iv. click on the particles you want to get rid of
    • v. when done, write the results to a new stack file
    • vi. delete input stack file.
      • Warning: deleted particles vanish permanently. However, assuming the micrographs are still around and one knows the parameters of the automated particle program in #3, the original stack files can be easily recreated.

  2. Convert individual stack files into one stack file that has to be in hdf format and should have the following attributes set (read more in I_O, examples are found in sparx/templates):

    • 1. For 2-D alignment: set alpha, sx, sy, and mirror (all to zero).

    • 2. For projection alignment (structure refinement): phi, theta, psi, s2x, and s2y have to be set in the header of each file. If their values are not known, all should be set to zero.

    • 3. CTF parameters (always): Pixel_size, defocus, voltage, Cs, amp_contrast, B_factor. In addition, if data was premultiplied by the CTF, ctf_applied has to be set to one, otherwise, set to zero. B_factor has default value of 0 and amplitude contrast 0.1. The defition of the latter may vary from one package to the other, for the definition used in SPARX, see filt_ctf.

  3. Run 2-D alignment programs, first sxali2d c, for high resolution work next sxali2d e.

  4. Run 2-D classification. First apply alignment parameters to the images in the stack file used in step 3 by running sxtransform2d and on the resulting stack run sxk_means and sxk_means_groups

  5. Determine the initial structure using class averages from step 4.
  6. Run 3-D projection alignment sxali3d d.

Cryo (last edited 2015-10-08 10:22:55 by penczek)