8. Extracting pseudosubtomograms

Doppio provides a wrapper around RELION’s pseudosubtomogram extraction tool and is called Extract subtomogram volumes job (relion.pseudosubtomo). A subtomogram is a small 3D volume extracted from a larger tomogram at the position of a particle of interest. However, RELION extracts them from the tilt series images and save each of them as a CTF-premultiplied stack of 2D images, called pseudosubtomograms. RELION can generate 3D subvolumes as well by combining the 2D images in the Fourier space, but working directly with 2D stacks is faster and space efficient. We refer to the RELION documentation for more details on pseudosubtomograms.

The job parameters we used for extracting pseudosubtomograms are shown in the image below.

Since the subtomograms are extracted from the tilt series images, the Binning factor does NOT need to match the tomogram binning factor. At this level the pixel size of the subtomogram is 1.35 x 6 = 8.1 A, while the tomogram pixel size is 10 A.

8.1. Practical tips for determining the Binning factor and box size for extraction

At the start, it is a good idea to extract particles at the same (or similar) binnig factor of the tomogram reconstruction. This will allow you to quickly test different parameters in the initial classification and alignment steps. Once you have a good idea of the particle size and quality of the data, you can re-extract the particles at a lower binning factor (higher resolution) for further processing.

The box size should be large enough to contain the particle of interest plus some margin to accommodate the CTF effects. A good rule of thumb is to choose a box size that is at least 2-3 times the size of the particle. For example, if the particle size is 200 A, a box size of 600 A would be a reasonable choice. If the particle shape is anisotropic, it’s a good idea to choose a box size that can accommodate the largest dimension of the particle.

RELION uses cropped boxes for downstream processing, so the cropped box size may be chosen ~1.2 - 2 times the particle size. Keep in mind that the computational cost increases with the cube of the box size, so smaller is cheaper. Also, consider the box size to be even, and if possible, multiple of 8 for optimal FFT performance.

If mask is used during alignment and classification, ensure that the mask fits well within the cropped box size to avoid edge artifacts. These are general guidelines, and the optimal parameters may vary depending on the specific dataset and research goals.

More details about the RELION pseudosubtomogram extraction job can be found in the RELION documentation below. Relion Pseudosubtomogram Extraction job documentation.