3. In the “Fiducial Model Generation” step, select “Use patch
tracking to make fiducial model.”
4. Change the patch size, for cryo-tomograms a larger patch size
is recommended. Note that you can use the “Use boundary
mode” option to only patch-track a subarea of the tile-series.
This is often used to avoid areas with few features or with
contamination.
5. The “Advanced” button at the bottom gives options to play
with filter parameters, trimming pixels and range of views, etc.
6. Check “Break contours into pieces” to perform robust fitting
with a tuning factor.
7. Click “Track Patches.”
8. After this is done, go to “Fine Alignment.” Make sure to check
“Do robust fitting with tuning factor” and turn on “Find
weights for contours, not point”. Click “Compute Alignment.” This will give you a mean residual in the log window.
9. Press “View/Edit Fiducial Model.” All the tracks will be shown
with overlapping segments. Tracks with obvious mistracking
can be deleted. If needed, go back to step 4 above to refine the
region used for tracking until you reach the desired alignment
quality.
4.3 Rapid Tilt-Series
Processing
4.3.1 Continuous tilting
data
Due to low signal-to-noise ratio (SNR) of individual images, some
processing is required to enhance the contrast for fiducial tracking.
One strategy we use is to sum neighboring images to enhance the
contrast of fiducials at each tilt angle (Fig. 5). This requires neighboring images to first be stretched by an amount equal to the
cosines of its tilt angle and of the tilt angle of the reference image,
followed by motion correction. These functions have been coded in
the Neighbor-enhance script, which requires the IMOD package.
1. Download the Neighbor-enhance script and the accompanying
rotmagstr program (https://github.com/chreifi/fasttilt) and
add both to your list of environment variables on your linux
machine.
2. Run the Neighbor-enhance script to gain normalize, correct
defects, remove deviant pixels, stretch, align, and sum images,
and perform a rough alignment of each summed image of the
tilt-series. Outputs are the raw data after gain normalization,
defect correction, and deviant pixel removal (.st), a roughly
aligned contrast-enhanced tilt-series (.preali file) and the transforms (.xf, .xg).
3. Run the eTomo software in the IMOD package, using the raw
tilt-series as input. Enter values for the tilt axis angle, start
angle, and an estimate of tilt increment.
104
Mohammed Kaplan et al.
tracking to make fiducial model.”
4. Change the patch size, for cryo-tomograms a larger patch size
is recommended. Note that you can use the “Use boundary
mode” option to only patch-track a subarea of the tile-series.
This is often used to avoid areas with few features or with
contamination.
5. The “Advanced” button at the bottom gives options to play
with filter parameters, trimming pixels and range of views, etc.
6. Check “Break contours into pieces” to perform robust fitting
with a tuning factor.
7. Click “Track Patches.”
8. After this is done, go to “Fine Alignment.” Make sure to check
“Do robust fitting with tuning factor” and turn on “Find
weights for contours, not point”. Click “Compute Alignment.” This will give you a mean residual in the log window.
9. Press “View/Edit Fiducial Model.” All the tracks will be shown
with overlapping segments. Tracks with obvious mistracking
can be deleted. If needed, go back to step 4 above to refine the
region used for tracking until you reach the desired alignment
quality.
4.3 Rapid Tilt-Series
Processing
4.3.1 Continuous tilting
data
Due to low signal-to-noise ratio (SNR) of individual images, some
processing is required to enhance the contrast for fiducial tracking.
One strategy we use is to sum neighboring images to enhance the
contrast of fiducials at each tilt angle (Fig. 5). This requires neighboring images to first be stretched by an amount equal to the
cosines of its tilt angle and of the tilt angle of the reference image,
followed by motion correction. These functions have been coded in
the Neighbor-enhance script, which requires the IMOD package.
1. Download the Neighbor-enhance script and the accompanying
rotmagstr program (https://github.com/chreifi/fasttilt) and
add both to your list of environment variables on your linux
machine.
2. Run the Neighbor-enhance script to gain normalize, correct
defects, remove deviant pixels, stretch, align, and sum images,
and perform a rough alignment of each summed image of the
tilt-series. Outputs are the raw data after gain normalization,
defect correction, and deviant pixel removal (.st), a roughly
aligned contrast-enhanced tilt-series (.preali file) and the transforms (.xf, .xg).
3. Run the eTomo software in the IMOD package, using the raw
tilt-series as input. Enter values for the tilt axis angle, start
angle, and an estimate of tilt increment.
104
Mohammed Kaplan et al.