This repository contains code for the paper: Sparse Fourier Backpropagation in Cryo-EM Reconstruction. Part of Advances in Ne 733B ural Information Processing Systems 35 (NeurIPS 2022) Main Conference Track.
You need to setup a Python environment with dependencies. We recommend installing via Miniconda3.
Once you have conda setup, you can install all the Python dependencies into a new environment by running:
conda env create -f environment.yml
You can then activate the conda environment by running:
conda activate sbackprop
Once inside the correct environment you can compile and install the CUDA dependencies by running:
python setup.py install
You can then run training by running
python voxelium/vae_volume/train.py <input STAR-file> <logdir> --gpu 0
Use -h for more options.
You can then visualize the results using
python voxelium/vae_volume/volume_explorer.py <logdir>
@article{kimanius2022sparse,
title={Sparse fourier backpropagation in cryo-em reconstruction},
author={Kimanius, Dari and Jamali, Kiarash and Scheres, Sjors},
journal={Advances in Neural Information Processing Systems},
volume={35},
pages={12395--12408},
year={2022}
}