Gnina - dkoes/docs GitHub Wiki

Gnina serves as the groups primary way to perform docking. There are 2 basic functionalities: Scoring with the CNN or the Autodock Vina scoring function. Using the CNN scoring function in the rescore option (the default setup) is the preferred way to use Gnina for docking. This wiki will explain a little extra about the results of running gnina --help and show some common use case scenarios

• Follow the instructions at https://github.com/gnina/gnina

• --score_only -- only using the scoring function to score the given pose (no docking)
• --minimize -- only uses the scoring function to go into the energy minimization subroutine on given poses (no docking)
• --cnn_scoring -- choose where to use the CNN scoring function in the molecular docking pipeline:
  • none -- does not use the CNN scoring anywhere within the docking pipeline
  • rescore -- (DEFAULT) uses CNN scoring for the final scoring and sorting of the ligand poses
  • refinement -- uses CNN scoring for ligand refinement and the final rescoring (relatively slow)
  • all -- uses CNN scoring at every point in the pipeline (extremely slow)

• using built-in CNN model: gnina -r <receptor structure> -l <ligand structure> --autobox_ligand <structure at the center of search> -o <output file>
• using different model + weights: gnina -r <receptor structure> -l <ligand structure> --autobox_ligand <structure at the center of search> -o <output file> --cnn_model <model file> --cnn_weights <weights file>

In the output file, the <CNNscore> is the score (0-1) assigned to the pose by the CNN, higher is better. <CNNaffinity> is the predicted affinity of this particular pose to the receptor structure, higher is better. <minimizedAffinity> is the score provided by AutoDock Vina, lower is better.

• gnina -r <receptor structure> -l <ligand structure> --autobox_ligand <structure at the center of search> -o <output file> --cnn_scoring none

In output file, the <minimizedAffinity> property of the molecule is the Vina score. More negative is better