|Tutorial - Tomographic Reconstructions in Priism (2/2)|
|7. Determine tilt axis select Determine tilt axis (if you already know this from other recons at same mag, this can be skipped). Note: the accuracy of this step is critically dependent upon the tilt offset being correct (eg the 0 degree tilt is actually 0 degrees). Mass normalization will usually do a good job of this, but you can do a better job when you view the back projection later. Iterating through this whole process will therefore give you the best possible answer.
A. Tilt series = file_b4.mrc (using the unbinned data might provide a better answer here if so, use the unbinned, but make sure your output file is still file_b4.bprmMn)
B. Search range = default is fine. You can play with this if your answers don't make sense.
C. First tilt pair = default (40.00 10.00) is fine for negative stain, for cryo I've found 30.00 0.00 useful.
D. Special parameters = more things to play with to refine your solution (don't use on first pass)
E. DoIt => file_b4.bprmMn (as always, overwriting the prior .bprmMn file)
8. Alignment select Alignment (we will set up this to EWBP and DoIt in one go) For Axis orientation, leave blank and the parameter determined above will be used, or enter best number (determined elsewhere) and check overwrite. This is the step where the X,Y translations to align the series are determined by cross correlation.
Note: If you are having alignment problems, check the box under special parameter to output the cross correlations and see what they look like. By changing the Weiner coefficient and 2nd peak threshold, you may be able to overcome some alignment problems. Turning on/off stretching, filtering, and phase weighting can also alter the output alignments, perhaps in a good way for some data sets. Below are examples of cross correlations at low and higher tilt angles (left and right respectively).
9. Apply parameters select Apply parameters. This step applies the translations as well as the normalization to the projected images to account for mass and the model of absorption assumed.
A. Aligned series = file_b4.MnAln (below)
B. Image formation = scaled linear. You can also select a logarithm or linear model. The linear model makes no changes in the densities.
C. special parameters, pcbase = 0.001 (tilt offset if you know it), apply alignment should be checked.
10. EWBP select EWBP = Elliptically weighted back projection. This is what creates your three dimensional file from the aligned stack. As a first pass, we generate only a central slice to be sure we have the tilt offset correct and the object of interest centered.
A. AlignedSeries = file_b4.MnAln (default)
B. Reconstruction = file_b4.xzyw (default)
C. Y range reconstructed = 250 250 1
D. DoIt => file_b4.bprmMn from alignment, file_b4.MnAln from application of alignment to tilt series, and file_b4.xzyw from EWBP.
11. Display EWBP reconstruction. Data should be contained within reconstruction and be horizontal. If data is not horizontal, note number of degrees from horizontal, clockwise rotation is positive. If data is not centered in reconstruction, note number of pixels from center using File Point Values from inside displayed window. Pixels above center are positive (see figure below).
12. Correct problems with centering and angles.
A. Uncheck Alignment
B. In Apply Parameters, change Output shift z value (third value) to the number of pixels from center observed above.
C. Special parameters - enter number of degrees from horizontal in Tilt offset
D. DoIt => new .MnAln and .xzyw. Note, display EWBP from a new CopyRegion window, otherwise the old image is held in the buffer and displayed
E. Look at your reconstruction as in step 12.
F. Adjust parameters as above until reconstruction is centered and horizontal.
G. Note extent of data in Z, you will use this for Iterative Tapir
H. Redo Alignment entering your final value for Tilt Offset in the special parameters.
I. Redo Apply parameters with no Tilt Offset. Keep your Output shifts.
J. DoIt => new .bprmMn and .MnAln files
13. EWBP (optional end point if not going the iterative route)
A. Uncheck Apply parameters
B. In EWBP change Y to reflect entire reconstruction (ie 0 499 1)
C. DoIt => .xzyw
D. Flip => .xyzw (Do this from command line Flip file.xzyw file.xyzw or through DataViews Rotation Flip).
E. Look at your reconstruction in both directions (xyzw, xzyw) using ViewFile or CopyRegion
14. Go to iterative reconstruction (Processing EM Processing Iter. Alignment + Reconstructions)
A. Tilt series = file_b4.mrc
B. InAlignParam = file_b4.bprmMn
C. OutParamBase = file_b4 (default)
D. Reconstruction = file_b4.xzy
E. Reconstruction shift = pixel value in z (third position) from step 12
F. Reconstruction Z size = extent of data in Z from step 12 (leave a bit of slop either side ~10% volume each side this hasnÕt been rigorously tested)
G. Number of runs = 10 (This seems good for me so far)
H. Special parameters, change pcbase = 0.001, remainder should be fine as default.
I. DoIt => series of Tapir Recon, Reproj, Align, Tapir = file_b4.xzy
J. Flip file_b4 => file_b4.xyz (look at this with either ViewFile or CopyRegion)
15. Stepping up to higher resolutions. At this point, if the reconstruction is good, you can proceed to higher resolution reconstructions.
Use the Alignment File Manipulator to change the last .bprmMn file from step 15 (ie file_b4.bprmMn_10) to a file for one step up in resolution (sampling ratio = 2, output file = file_b2.bprmMn).
Check your reconstruction z shift settings as in step 13. This can change substantially during iterative Tapir. (The input tilt series is now file_tb2.mrc, or file_t.mrc)
Setup iterative Tapir as above. I usually do 5 runs here.
If this goes well, proceed to a full resolution reconstruction.
The dataset included has some really nice structure in the centriole, despite the noise. Below are a Butterworth filtered view of the bin2 reconstruction done with Filter3D(Fourier) and a projection view of that same volume surrounding the centriole done with Volume Viewer.
|Return to page 1.|