Generate point cloud

Use POINTNET files to generate point cloud data for proteins and ligands.All data must be in PDB format. All data should remove solvents, metals and ions. And use openbabel to add polar hydrogen (--AddPolarH)

./POINTNET {protein_path} {ligand_path} {out_path}
./POINTNET-2048 {protein_path} {ligand_path} {out_path}
./POINTNET-atomchannel {protein_path} {ligand_path} {out_path}

Prediction

  1. Installation dependent environment
 conda create -n point_cloud_envs
 conda activate point_cloud_envs
 conda install python=3.7
 conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge
  1. Prediction
    python pred.py --file ./example/5c2h_11.09 --model PointTransformer

Other model parameters can be downloaded here

Machine Learn data

The machine learning script, as well as the training and test data, are available via the URL below.

Feature and script here

model for extract feature here

PointCloud Data

Download the PDBBind all data and Bigdata point cloud data using the URL below.

PDBBind all data here