Fig. 2: Optimization of the search parameters in template matching.

a–c Dependence of the average constrained cross-correlation peak height (z-scores) for 80S ribosomes, NPC subunit, half vault, and microtubule on the high-pass (a) and low-pass filters (b), and on the diameter of the spherical mask for the ribosome and a cylindrical mask for the microtubule (c). In a no low-pass filter was applied, and in b no high-pass filter. d Schematic representation of the overlapping voxels (orange) when a microtubule is rotated around its optimal orientation by 20 deg. e, f Dependence of the constrained cross-correlation of a template with itself (in silico evaluation) as a function of the fraction of overlapping voxels and the angular distance (f). e, f present cross-correlation coefficients (and not z-scores) for the template matched against itself, not the tomogram. In e the CC values for all the rotations in a 10 degrees grid are shown, in f the average of the CC values for all rotations sampled in a 1-degree grid and binned every 2 degrees. For the average, the number n increases as a function of the angular distance (from n = 4 at the pole to up to n = 306 at 50 deg). In all cases, error bars correspond to one standard deviation. In a–c n = 10, 45, 4, and 16 for NPC subunit, ribosome 80S, half vault, and microtubule, respectively. Source data are provided as a Source Data file.