Sub-tomogram averaging

The following was written by Tanmay Bharat.

Getting started

Tomograms should be in .mrc format. Coordinates for the sub-tomograms should be in 3-column ASCII files with X Y Z coordinates.

Generating 3D CTF volumes for sub-tomogram averaging

• First CTFFIND should be run for all images of the tilt series (that was used to generate the tomogram). The results should be written out in a text file with the syntax:

Tilt_angle average_defocus_value_from_CTFFIND

This can be done automatically using the script run_ctffind.py [TO BE ADDED]. Please modify the header values of the script.

• Now the coordinates from the text file and the average defocus values from CTFFIND will be used to calculate local CTF parameters for each sub-tomogram in each image of the tilt series. This can be done using the

script make_ctfstar.py [TO BE ADDED]. Please modify the header values in this python script. For each subtomogram, a RELION .star file will be written out. This will be used by relion_reconstruct to generate the 3D CTF volume.

• Run relion_reconstruct to generate all 3D CTF model volumes. This can be done using the script reconstruct_CTF_volumes.sh that was written out in the last step. This script uses the .star file to read the local CTF

parameters and use them to make the 3D CTF model volume.

Sub-tomogram extraction

• Extract sub-tomograms using RELION as you would do for 2D micrographs. If the image file is a volume and the coordinates values contain X,Y,Z entries, automatically sub-tomograms will be written out rather than 2D particles. If you add --project3d to the RELION extraction command, then 2D projections of each sub-tomogram will be written out. This may be useful to run reference-free 2D-class averaging in order to identify junk or otherwise unsuitable particles in your data set.

• Run "3D Auto-Refine" or "[Classify_3D_structural_heterogeneity | 3D classification]" jobs, just like you would for single particles specifying the CTF volume for each image in a .star file. This can be done using the

script prepare_subtomo.py. A sample output is shown in the script.