TankBind could predict both the protein-ligand binding structure and their affinity. If you have any question or suggestion, please feel free to open an issue or email me at wei.lu@galixir.com or shuangjia zheng at shuangjia.zheng@galixir.com.
conda create -n tankbind_py38 python=3.8
conda activate tankbind_py38
You might want to change the cudatoolkit version based on the GPU you are using.:
conda install pytorch cudatoolkit=11.3 -c pytorch
conda install torchdrug=0.1.2 pyg biopython nglview jupyterlab -c milagraph -c conda-forge -c pytorch -c pyg
pip install torchmetrics tqdm mlcrate pyarrow
p2rank v2.3 could be downloaded from here:
https://github.com/rdk/p2rank/releases/download/2.3/p2rank_2.3.tar.gz
We use the prediction of the structure of protein ABL1 in complex with two drugs, Imatinib and compound6 (PDB: 6HD6) as an example for predicting the drug-protein binding structure.
examples/prediction_example_using_PDB_6hd6.ipynb
Scripts for training dataset construction will be released later.
TankBind also support virtual screening. In our example here, for the WDR domain of LRRK2 protein, we can screen 10,000 drug candidates in 2 minutes (or 1M in around 3 hours) with a single GPU. Check out
examples/high_throughput_virtual_screening_LRRK2_WDR.ipynb