Learning Structure-based Subpocket Representations for Protein-Ligand Interaction Prediction
cuda 11.2
python 3.7.4
torch 1.7.1
torch-geometric 1.6.3
numpy 1.19.0
pandas 1.2.4
rdkit 2020.03.3.0
scikit-learn 0.21.3
scipy 1.6.3
tensorboard 2.4.1
-
Prepare a environment that satisfying the above requirements;
-
Download the trained model files:
-
Download the input data files:
-
Run the inference scripts below;
-
The results can be found in
[TASK]/results/[FOLDER]/
.
Protein ligand binding site prediction
python runPrediction.py --task PocketDetection --dataset COACH420
python runPrediction.py --task PocketDetection --dataset HOLO4k
Protein-ligand binding affinity prediction
python runPrediction.py --task Affinity --setting original
python runPrediction.py --task Affinity --setting newprotein
python runPrediction.py --task Affinity --setting expanded
Non-covalent interaction prediction
python runPrediction.py --task Interaction
Protein ligand binding pose classfication
python runPrediction.py --task Distance