The dataset is intended for the pytom-gpu-template-matching module for running GPU-accelerated template matching. Details of the method can be found in Chaillet et al., 2023.

The dataset contains 10 tomographic reconstruction from cryo-electron microscopy tilt-series. The reconstruction were generated as follows: (1) MotionCor2 was used to correct the motion in raw movie frames, (2) ctfplotter from IMOD was used to estimate defocus values, (3) ctfphaseflip from IMOD was used for strip-based phaseflipping, and (4) AreTomo was used to align and reconstruct tomograms (weighted back-projection) with 5x5 local patches. Files are saved in the MRC format and the name has the structure tomo[DATE]_[ID].mrc, where DATE refers to collection date and ID refers to the tilt-series index.

Each tomogram additionally contains three files using the same name format, but ending in .rawtlt, .defocus, and _dose.txt. These files contain tilt-angles used for collecting the tilt-series (in degrees), accumulated dose per tilt (in e-/A2), and an IMOD-style defocus files with estimated defocus per tilt, respectively.

Each tomogram also contains a STAR file with manually inspected and corrected annotations, resembling some sort of a 'ground truth' annotation. The files end with *_bin_manual.star and are in the Relion4 STAR format.

The tutorial and software can be found on Github:

tutorial instructions download software

The raw tilt-series can be found in this DataverseNL repository doi:10.34894/OLYEFI, they were collected as described in Gemmer et al., 2023 for which there is an EMPIAR dataset available EMPIAR-11751.

MotionCor2, 1.5.0

AreTomo, 1.3.3