IMOD

Overview

IMOD is an EM tomography software suite that has been developed by the Boulder Laboratory. Although basically runs through command line interfaces and scripts, most users learn to use IMOD using its GUI called eTomo.

Contents

Using IMOD

Things you will need:

A tilt series aka your dataset acquired on an EM
A .rawtlt file or knowledge of the tilt angles that were acquired
A computer that has IMOD installed (NCMIR personnel click here for list of machines to log into)

Things that are helpful:

Size of gold fiducial markers
Magnification

The Boulder Laboratory has an easy to follow tutorial found here and thus, no step-by-step instruction will be provided on these pages.

NCMIR STAFF: Please remember not to download the tutorial dataset to your home directory but instead to one of the network drives. See a network admin or your supervisor where data should be saved.

Useful IMOD hotkeys

No information at this time

Conceptual steps of IMOD backprojection

Making the .rawtlt file
Pre-processing

Coarse Alignment
Fiducial Model Generation
Fine Alignment
Tomogram Positioning
Final Aligned Stack
Tomogram Generation
If reconstructing a double-tilt dataset, the next few steps apply:
Tomogram Combination
Post-processing
Clean Up

IMOD files: what they all mean

FAQ

How do I learn to use IMOD?
- The IMOD homepage hosted at Boulder Laboratory has an excellent tutorial and is constantly updated as new versions of IMOD is released. The link to the tutorial is http://bio3d.colorado.edu/imod/doc/etomoTutorial.html
I type etomo in a terminal window but it won't start up!
- If you're on a Linux machine, did you remember to type imod_setup before typing etomo? You can avoid typing this each time by adding these lines to your .cshrc file:
```
 setenv IMOD_DIR [directory path to imod/imod_[version number]]
 source /[directory path to imod]/IMOD-linux.csh\
```
Can I use IMOD on my home computer?
- Yes. Download and install IMOD from the IMOD homepage, or remotely login to your lab machine using ssh or nxclient. Note: depending on the size and location of your dataset, along with the speed of your home machine, it may be faster to use IMOD from your work / lab.
The contrast on my .preali file looks strange or blown out!

An example of a "blown-out" preali view
- This is sometimes just a display error and nothing wrong with your dataset per se. Often times occurs if the dynamic range is narrow. If this is bothersome, try reloading your dataset by:
  - go to Edit – > Image – > Reload... in 3dmod Control Window
  - Typically the given limits are fine and clicking on "Apply" will reload the display with new limits.
  - If the reloaded display is undesirable, you can close the 3dmod window and re-open the file. Reloaded changes are not saved.
How many gold fiducials should I try to track?
- The complicated answer is "depends". Shoot for somewhere between 10 - 100. Keep in mind that the more fiducials you choose that are distributed across the xy and on both sides of the sample, you will generally get a better alignment. Anything more than 100 will see a giant leap in the cost vs benefit ratio of time to alignment (read: takes too long to track).
The fiducials aren't tracking from the .seed file! Why?!
- Possible causes:
  1. Correlation errors between the .rawtlt file (or tilt header) and the images
  2. Image rotation number was incorrect when setting up your .edf file. Double check this value by looking in the .edf file .
  3. Change tracking parameter values. Perhaps the entered diameter of gold is wrong? Note: when in doubt, choose a smaller diameter value than a larger one. This is particularly useful to know if there are more than one sized gold fiducial marker.
Help! I can't get my residual error to the levels the tutorial had! What's an "acceptable" residual error level?
- Depends on several factors. Don't be surprised if your "acceptable" residual error level goes up with the xy size of dataset, etc.
What can I do to lower my residual error?
1. Add more fiducials
2. Remove troublesome fiducials / ambiguous fiducials with unusually high residual errors
3. Run through the Fiducial Alignment step several times with lower thresholds for reporting
4. Double check that you have the correct number of surfaces for analysis in fine alignment. Do you really have gold on both sides if the "2 surfaces for analysis" radio button is depressed?
How can I produce a better tomo?
- Try the following:
  1. Increase the number of tilt series e.g. single vs multiple.
  2. Increase the number of angular increments in each tilt series and/or expand the tilt range from the typical +/- 60 degrees e.g. +/- 70 degrees
  3. Make sure that the raw data is in focus. Check for blurriness, drift, poor signal:noise and throw out any tilt that are deemed unuseable.
  4. Try different filter settings
  5. Place more / fewer fiducials
  6. Double check the accuracy of your tomopitch

Common Error Messages