gninatorch.gnina module

gninatorch.gnina.load_gnina_model(gnina_model: str, dimension: float = 23.5, resolution: float = 0.5)[source]

Load GNINA model.

Parameters:

  • gnina_model (str) – GNINA model name

  • dimension (float) – Grid dimension (in Angstrom)

  • resolution (float) – Grid resolution (in Angstrom)

gninatorch.gnina.load_gnina_models(model_names: Iterable[str], dimension: float = 23.5, resolution: float = 0.5)[source]

Load GNINA models.

Parameters:

model_names (Iterable[str]) – List of GNINA model names

gninatorch.gnina.main(args)[source]

Run inference with GNINA pre-trained models.

Parameters:

args (Namespace) – Parsed command line arguments

Notes

Models are used in evaluation mode, which is essential for the dense models since they use batch normalisation.

gninatorch.gnina.options(args: List[str] | None = None)[source]

Define options and parse arguments.

Parameters:

args (Optional[List[str]]) – List of command line arguments

gninatorch.gnina.setup_gnina_model(cnn: str = 'default', dimension: float = 23.5, resolution: float = 0.5) → Module | bool[source]

Load model or ensemble of models.

Parameters:

  • cnn (str) – CNN model name

  • dimension (float) – Grid dimension

  • resolution (float) – Grid resolution

Returns:

Model or ensemble of models

Return type:

nn.Module

Notes

Mimicks GNINA CLI. The model is returned in evaluation mode. This is essential to use the dense model correctly (due to the nn.BatchNorm layers).