This is code accompanying the paper, https://arxiv.org/abs/2401.15183.

We provide 4 scripts: demo_synthetic_data.py, demo_experimental_data.py, precompute_from_pdb.py, precompute_from_map.py

A list of requirements are stored in requirements.txt, which can be installed via the command pip install -r requirements.txt.

To run the code, one must first generate Clebsch-Gordan coefficients by running utils/generate_CG_coefficients.py. Then one can create least squares matrices using the appropriate precompute scripts. Here, we use precompute_from_pdb.py to generate the data needed for demo_synthetic_data.py and precompute_from_map.py to generate the data needed for demo_experimental_data. The directory to save/read data can be changed by editing the save_path variable in each script.

Once the precompute steps are done, the demos are ready to run. Note that demo_experimental_data requires the particle stack downloaded from https://www.ebi.ac.uk/empiar/EMPIAR-10076/. For ease of use, one can download the images to the EMPIAR-10076/ folder in this directory. Further note that demo_volumes.py requires moment information to be stored from the precompute steps, which can be accomplished by uncommenting the lines 119-122 in precompute_from_map.py or 132-133 in precompute_from_pdb.py.

The code in utils/fast_cryo_pca.py, utils/utils_cwf_fast_batch.py, and utils/fle_2d_single.py are cloned from https://github.com/yunpeng-shi/fast-cryoEM-PCA/tree/main with minor changes to include more parameters.