Methods to reconstruct 3D structure from cryoEM data using adversarial learning [1].
Run following to install the dependencies:
conda env create --file environment-mac.yml (on mac)
conda env create --file environment.yml (on non-mac)
To start a reconstruction, run the following.
python main_cryogan.py --config_path configs/betagal_simulated.yaml
The parameters of the reconstruction can be edited in the input config file. For simualted the SNR of the ground truth data can also be given with:
python main_cryogan.py --config_path configs/betagal_simulated.yaml --snr_val -5
[1] Gupta H, McCann MT, Donati L, Unser M. CryoGAN: a new reconstruction paradigm for single-particle cryo-EM via deep adversarial learning. IEEE Transactions on Computational Imaging. 2021 Jul 13;7:759-74. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9483649