Alignment of Densities

In the following we will use template matching to recover the correct orientation of a density map within another density map.

To demonstrate the procedure, we will use EMD-15271. You can download it from the EMDB or from the command line

wget https://ftp.ebi.ac.uk/pub/databases/emdb/structures/EMD-15271/map/emd_15271.map.gz

We will first simulate a translation and rotation of the map

import numpy as np
from tme import Density
from tme.matching_utils import get_rotation_matrices

rotation_matrix = get_rotation_matrices(40)[32]

density = Density.from_file("emd_15271.map.gz")
density, _ = density.centered(0) # Create a sufficiently sized volume for rotation
density_mod = density.rigid_transform(
    rotation_matrix = rotation_matrix
)
density_mod.origin = np.add(
    density.origin, np.multiply((-10, 5, 0), density.sampling_rate)
)

density.to_file("emd_15271.mrc")
density_mod.to_file("emd_15271_mod.mrc")

The following will recover the correct orientation between the densities

match_template.py   \
    -m emd_15271.mrc \
    -i emd_15271_mod.mrc \
    -a 40 \
    -n 4 \
    --no_centering

The orientation with highest score, which in our case is the initial correct orientation, can be obtained from the created pickle files like so

postprocess.py \
    --input_file output.pickle \
    --number_of_peaks 1 \
    --output_format alignment \
    --output_prefix emd_15271_fit

The aligned densities are displayed below. The left side shows the map and emd_15271_mod.mrc, the right side the output of pytme.