☰
Getting Started
IMOD Processing Pipeline
Motion Correction
Build Tomogram / Setup
X-ray Removal
▶ Interactive: Coarse Alignment
Coarse Alignment
▶ Interactive: Patch Tracking
Alignment Model — Fiducial Model
▶ Interactive: Fine Alignment
Fine Alignment
Tomogram Positioning
CTF Estimation
▶ Interactive: CTF & Phase Contrast
Tomogram Reconstruction
CryoCARE Denoising
▶ Interactive: CryoCARE Explained
Getting Started
0. Open Alignframes GUI
Open Cygwin64 terminal (on desktop) and go where you copied the IMOD data folder
cd /cygdrive/c/Users/YOUR_USERNAME/Documents/IMOD/
Open etomo front page by typing:
etomo
IMOD Processing Pipeline
1. Motion Correction
2. Alignment
3. CTF correction
4. Tomogram reconstruction
Pre-alignment
CTF estimation
Tilt/Positioning
Post-processing
IMOD Processing pipeline
Motion Correction
~100𝑛𝑚
~100𝑛𝑚
Each frame of a tilt series is
composed of “sub-frames”
Motion Correction
𝚰 – IMOD processing – Correcting for motion artifacts
Two main forces of motion during data collection: stage drift and beam-induced ice deformation
Motion during sub-frame acquisition
Average of motion
Without motion correction
With motion correction
𝚰 – IMOD processing – Correcting for motion artifacts
Single “sub-frames”
1. Select the .mdoc
Motion Correction
2. Select the .tif images
Motion Correction
3. re-select “metadata” box
Motion Correction
4. Tick “Gain normalize”
box and select the .dm4
gain file
Motion Correction
5. Double check the other
parameters are similar and
go to Alignment tab
Motion Correction
6. Select “Advanced”
Motion Correction
7. Set Binning to 8
for alignment and 2
for output
Motion Correction
8. Check “Do dose
weighting” and “Dose
weighting from .mdoc
file
9. Check all parameters are
similar and Run Alignframes
Motion Correction
“tomo-g2_af.mrc”
has
been
created, view it in 3dmod with
“Open Output Tilt Series”
Or through the terminal:
3dmod -I 1 tomo-g2_af.mrc
(capital i )
Right click on the image to view the tilt series
Motion Correction
Build Tomogram / Setup
Build Tomogram
Open etomo front page again by typing in Cygwin terminal:
etomo
1. Select the tilt series
we motion corrected
3. Scan header and
enter “0” for “Fiducial
diameter (nm)”
2. Select “Single axis”
5. Check all parameters
are similar and “Create
Com Scripts”
4. Un-check “Series was..”
Setup
Equivalent CLI command:
newstack \
-InputFile tomo-g2_af.mrc \
-OutputFile tomo-g2_af_ali.mrc \
-TransformFile tomo-g2_af.xf \
-TaperAtFill 1,0 \
-AdjustOrigin \
-SizeToOutputInXandY 4092,20 \
-OffsetsInXandY 0,0 \
-ImagesAreBinned 0.5
Setup — .com files
Command files (C-shell scripts) that drive IMOD processing steps. One .com per step — newst.com, tilt.com, eraser.com, etc.
Structure — 3 types of lines
# → comment
$ → shell command ($setenv, $newstack, $if)
Everything else → keyword + value fed to the program via stdin
X-ray Removal
1. Select “Pre-
processing”
2 “Find X-rays”
and “View X-ray”
X-ray removal
1. “Create Fixed Stack”
2. “Use Fixed Stack”
3. “Done”
X-ray removal
Interactive: Coarse Alignment
Quick reference for coarse alignment
Coarse Alignment
1. “Calculate Cross-
Correlation”
2. Set binning to 4
3. “Generate Coarse
Aligned Stack” and
inspect tilt series
4. “Done”
Alignment model
Coarse Alignment
Interactive: Patch Tracking
Quick reference for patch tracking
Alignment Model — Fiducial Model
2.
Set
“Size
of
patches” to 300,300
3. Set patch overlap
to (0.8,0.8)
4.
Check
“Use
boundary model”
and create one
1. Select “Use patch
tracking”
Alignment model
Fiducial model
1. Select “Model” mode
2. Using mouse middle
button to draw boundary
around the cell
3. Press “s” to save, and
close windows
Alignment model
Fiducial model – Boundary model
“Track
Patches”
and
“Open Tracked Patches”
Alignment model
Fiducial model
Press “v” to open “Model View”
and inspect patch trajectories
Inspect patches in movie mode
Alignment model
Fiducial model
When using too many/small patches
Patch residuals vectors
Patch trajectories
Alignment model
Fiducial model
Patch residuals vectors
Patch trajectories
Without using boundary model
Alignment model
Fiducial model
“Done”
Alignment model
Fiducial model
Interactive: Fine Alignment
Quick reference for fine alignment (tiltalign)
Fine Alignment
1. Check “robust fitting”
2. Go to “Global Variables” tab
Alignment model
Fine Alignment
Set Rotation, Magnification,
Tilt Angle, and Distortion
solution types
Alignment model
Fine Alignment
1. Run cross-validation
2. Compute Alignment
Alignment model
Fine Alignment
“View 3D model”
“View residual vectors”
Alignment model
Fine Alignment
Tomogram Positioning
Tomogram positioning & Volume trimming
X
Y
X
Z
Y=1/4
Y=2/4
Y=3/4
Prompts user to delineate tomogram volume
Y=1/4
Y=2/4
Y=3/4
1. Set tomogram
thickness to 3000
2. Check “Use whole
tomogram” and use
“Binning 8”
3. Run “Create Whole
Tomogram” and “Create
Boundary Model
Tomogram positioning & Volume trimming
1. Rotate the tilt series to view the Z/X view
2. Set Z value to roughly 1/3rd the thickness
3. (in “Model” mode) use middle-click to set 2 pairs
of points on either side of the biological sample
Tomogram positioning & Volume trimming
4. Repeat for Z ≈ 2/4 and Z ≈ 3/4
5. Press “s” to save and close
Tomogram positioning & Volume trimming
Opening XYZ view can be helpful
1. Run “Compute Z
shift” and “Create
Final Alignment”
2. “Done”
Tomogram positioning & Volume trimming
CTF Estimation
CTF estimation
“Thon” rings visible in Power
spectrum (Fourier transform)
Defocus and microscope abherations introduce spatial frequency (phase) distorsions.
Contrast Transfer Function (CTF) needs to be approximated to recover true object
Interactive: CTF & Phase Contrast
Explore defocus, phase contrast, and the contrast transfer function
1. Set Binning to 4
and run “Create
Full Aligned Stack”
CTF estimation
1. Click “View Full Aligned Stack”
2. In 3dmod press Shift+F to view tilt series in Fourier space
CTF estimation
1. Click “Correct CTF” tab
CTF estimation
1. Click “View Full Aligned Stack”
CTF estimation
4. “Scan for Defocus then Autotune”
Set “lower end of defocus range” to 2
and “higher end” to 8”
3. Check “Vary exponent of CTF function”
Check “Do auto weighing and truncation”
6. “Save to File”
5. “Autofit All Views”
1. Set “Autofit” from -60 to 60
2. Check “Fit each view seperately”
CTF estimation
1. Click “Correct CTF”, “View CTF Correction”
and “Use CTF Correction”
CTF estimation
Before CTF correction
After CTF correction
Tomogram Reconstruction
Tomogram Reconstruction
“Generate Tomogram”
Tomogram Reconstruction
1. Check “Rotate around X axis
2. “Trim Volume”
We can reorient the tomogram if it’s in the wrong orientation
Tomogram Reconstruction
CryoCARE Denoising
~100𝑛𝑚
“Sub-frames” are separated into “even”
and “odd” stacks
CryoCARE denoising
Preparing “Even/Odd” stacks
~100𝑛𝑚
“Even” motion corrected frame
“Odd” motion corrected frame
CryoCARE denoising
Preparing “Even/Odd” stacks
alignframes -mdoc s1.st.mdoc -output tomo-g2_af.mrc -adjust -bin 8,2 -gain SuperRef_tomo-g2_00000_-11.0.dm4
-pixel 1.283 -evenodd 1
Creates “tomo-g2_af_even.mrc”
and “tomo-g2_af_odd.mrc”
Use “combined”
transformation files
Use “combined”
transformation files
We apply the same transformation files to “Even” and “Odd” frames to generate tomograms that are
identical in structure but differ only in the noise.
Train &
denoise
~2 hour
~10 mn
~10 mn
Odd Tomogram
Even Tomogram
CryoCARE denoising
Preparing “Even/Odd” stacks
Denoised Tomogram
CryoCARE denoising
Reconstructing “Even/Odd” tomograms
mkdir even
mkdir odd
move tomo-g2_af_even.mrc even
move tomo-g2_af_odd.mrc odd
copy tomo-g2_af.xf even
copy tomo-g2_af.xf odd
copy tomo-g2_af.tlt even
copy tomo-g2_af.tlt odd
1. Create folders and move even/odd stacks
2. Copy alignment files into both folders
cd even
newstack -input tomo-g2_af_even.mrc -output tomo-g2_af_even_ali.mrc
-xform tomo-g2_af.xf -taper 1,0 -origin -size 1024,1440 -bin 4
3. Align even stack (newstack)
4. Reconstruct even tomogram (tilt)
tilt -input tomo-g2_af_even_ali.mrc -output tomo-g2_af_even_rec.mrc
-IMAGEBINNED 4 -TILTFILE tomo-g2_af.tlt -THICKNESS 1800 -RADIAL
0.35,0.035 -FalloffIsTrueSigma 1 -LOG 0.0 -SCALE 0.0,250.0 -MODE 1 -
AdjustOrigin
6. Do the same in odd folder (replace the _even_ by _odd_ in
the commands)
clip rotx tomo-g2_af_even_rec.mrc tomo-g2_af_even_rotx.mrc
5. Rotate even tomogram (clip)
train_data_config.json
train_config.json
predict_config.json
CryoCARE training and prediction is controlled through 3 configuration files and 3 command lines
CryoCARE denoising
Runnig CryoCARE
~30 mn training
20 epochs / 75iterations
~6 hours training
200 epochs / 100 iterations
Original
CryoCARE denoising
Interactive: CryoCARE Explained
Understand how CryoCARE denoising works