deepmol.utils package
Submodules
deepmol.utils.cached_properties module
deepmol.utils.decorators module
- inplace_decorator(method: callable) callable | None [source]
Decorator to make a method inplace.
- Parameters:
method (callable) – Method to decorate.
- Returns:
inplace_method – Decorated method.
- Return type:
callable
- modify_object_inplace_decorator(method: callable) callable | None [source]
Decorator to create a lazy copy-on-write version of a method.
This decorator performs modifications of an object that is received by the class, either inplace or on a copy of the object, depending on the value of the inplace parameter.
This applies inplace
- Parameters:
method (callable) – The method to decorate.
- Returns:
modify_object_wrapper – The decorated method.
- Return type:
callable
deepmol.utils.errors module
deepmol.utils.utils module
- canonicalize_mol_object(mol_object: Mol) Mol [source]
Canonicalize a molecule object.
- Parameters:
mol_object (Mol) – Molecule object to canonicalize.
- Returns:
Canonicalized molecule object.
- Return type:
Mol
- get_class(name: str) object [source]
Get a class from a string.
- Parameters:
name (str) – A class name you want to get.
- Returns:
A class object.
- Return type:
object
- load_from_disk(filename: str) Any [source]
Load object from file.
- Parameters:
filename (str) – A filename you want to load.
- Returns:
A loaded object from file.
- Return type:
Any
- load_pickle_file(input_file: str) Any [source]
Load from single, possibly gzipped, pickle file.
- Parameters:
input_file (str) – The filename of pickle file. This function can load from gzipped pickle file like XXXX.pkl.gz.
- Returns:
The object which is loaded from the pickle file.
- Return type:
Any
- mol_to_smiles(mol: Mol, **kwargs) str | None [source]
Convert SMILES to RDKit molecule object. Parameters ———- mol: Mol
RDKit molecule object to convert.
- kwargs:
Keyword arguments for rdkit.Chem.MolToSmiles.
- smiles: str
SMILES string.
- normalize_labels_shape(y_pred: List | ndarray, n_tasks: int) ndarray [source]
Function to transform output from predict_proba (prob(0) prob(1)) to predict format (0 or 1).
- Parameters:
y_pred (array) – array with predictions
n_tasks (int) – number of tasks
- Returns:
Array of predictions in the format [0, 1, 0, …]/[[0, 1, 0, …], [0, 1, 1, …], …]
- Return type:
labels