torch-reconstruct-tomogram

(sub-)Tomogram reconstruction and subtilt extraction for cryo-ET.

Overview

This package provides (sub-)tomogram reconstruction and subtilt extraction driven entirely from a torch-tilt-series TiltSeries. It supports

• extract_particle_tilt_series(): extract a subtilt-series at 3D location(s) in the sample
• reconstruct_subvolume(): rank-polymorphic reconstruction of 3D patch(es) at location(s) in the sample
• reconstruct_tomogram(): full volume reconstruction by tiling reconstructed patches in 3D

TiltSeries holds alignment geometry (in Angstroms) plus image_path/image_indices/pixel_spacing metadata. The functions above take a TiltSeries, and load and (by default) preprocess the matching raw images internally via torch_tilt_series.load_tilt_series_images() / preprocess_tilt_series_images() (by default: plane subtraction, a DC-excluding bandpass with no low-pass, i.e. up to Nyquist, and central-crop normalization); pass **preprocessing_kwargs to override any of its defaults (low=, high=, falloff=, bandpass_padding=, subtract_background=, normalize=) - see preprocess_tilt_series_images()'s docstring for details. output_pixel_spacing lets both local (reconstruct_subvolume) and global (reconstruct_tomogram) reconstruction target an arbitrary output voxel size. Reconstruction happens at the input pixel spacing and is Fourier-rescaled to the requested output size. Reconstruction is performed in Fourier space using central slice insertion. Positions are in zyx coordinates, in Angstroms, relative to the tomogram center.

Installation

pip install torch-reconstruct-tomogram

To load a tilt series from AreTomo or ETOMO output, also install the IO dependencies for torch-tilt-series:

pip install torch-tilt-series[io]

Examples

See the examples/ folder for scripts showing how to load a tilt series, reconstruct subvolumes and tomograms, and save the result.

License

This project is licensed under the BSD 3-Clause License - see the LICENSE file for details.