Struc2mapGAN is a novel data-driven method that employs a generative adversarial network (GAN) with a U-Net++ architecture as the generator to produce improved experimental-like density maps from molecular structures (PDB files).
Python 3 :
https://www.python.org/downloads/
UCSF ChimeraX :
https://www.cgl.ucsf.edu/chimerax/download.html
numpy<2.0
torch==2.2.2
lightning==2.2.3
# Create conda environment
conda create -n struc2mapGAN python=3.10
conda activate struc2mapGAN
# Clone git repo
git clone https://github.com/chenwei-zhang/struc2mapGAN.git
cd struc2mapGAN
# Install dependencies
pip install -r requirements.txt# Download the checkpoint and save to ./ckpt/
wget --content-disposition -P ./ckpt https://osf.io/download/397v2/cd struc2mapGAN/app
# Generate maps
python struc2mapGAN.py --pdb ../example/8i2h_ref.pdb --ckpt ../ckpt/struc2mapGAN.ckpt --output_mrc ../example/8i2h_struc2mapGAN.mrc
# If resample to experimental map's box size
python struc2mapGAN.py --pdb ../example/8i2h_ref.pdb --ckpt ../ckpt/struc2mapGAN.ckpt --output_mrc ../example/8i2h_struc2mapGAN.mrc --ref_map ../example/emd_35136.map- --pdb <pdb_path> Path to the input pdb file
- --ckpt <ckpt_path> Path to the trained checkpoint
- --output_mrc <output_path> Path to the output map
- --ref_map <map_path> Path to the reference map file if resample to the original experimental map's box size
Please visit the OSF space at https://osf.io/zcxvs/ to view all the test maps generated using struc2mapGAN.
Chenwei Zhang (cwzhang@cs.ubc.ca)
This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details.