AQDnet: Deep Neural Network for Protein-Ligand Docking Simulation.

• aqdnet.py
  • Code that provides an interface to AQDnet's feature extraction.
• lpcomp.py
  • Code containing AQDnet's algorithm for feature extraction.
• model.py
  • Code that provides an interface for training models with AQDnet features.
• structure.py
  • Code that specifies the structure of AQDnet's deep learning model and its accompanying preprocessing methods..
• runner.py
  • Code that includes some utility functions such as feature loading..
• predict.py
  • Code that provides an interface to make inferences using AQDnet's trained models.

1. Ex1_generate_feature.ipynb
   • Feature generation example.
2. Ex2_train_model.ipynb
   • Model training example.
3. Ex3_predict.ipynb
   • Prediction example.

• Docking_Energy30RMSD2.5
  • Best model of AQDnet's Docking-specific model.
• Scoring_Energy02RMSD2.0
  • Best model of AQDnet's Scoring-specific model.

• Docking_Energy30RMSD2.5
  • Evaluation results of AQDnet's Docking-specific model with CASF-2016.
• Docking_AQDnet_summary.csv
  • Docking power test result of three different energy filtering conditions.
• Scoring_Energy02RMSD2.0
  • Evaluation results of AQDnet's Scoring-specific model with CASF-2016.
• ScoringPower_result.csv
  • the AQDnetet's result of Scoring power test and those of the other SFs.
• LIT-PCBA_result.csv
  • All tha result of the AQDnet's LIT-PCBA evaluation. EF1% of all template PDB ids are described.
• LIT-PCBA_result_summary.csv
  • Summarized LIT-PCBA result. Max, min, mean and SD of EF1% of all targets are descibed.

• Sample AQDnet features. Due to file size, only features for 5 complexes are available here.

Sample structures of 5 complexes.