data_models.descriptor_generators.api

data_models.descriptor_generators.api

Classes

Name Description
DescriptorGenerator Base class for all descriptor generators.
Fingerprints
Fragments
MordredDescriptors

DescriptorGenerator

data_models.descriptor_generators.api.DescriptorGenerator()

Base class for all descriptor generators.

A DescriptorGenerator is a pure, stateless transform: given a series of structure identifiers (e.g. SMILES) it returns a dataframe of numeric descriptor columns. Correlation-based decorrelation is not done here — it lives on the consuming descriptor encoding / engineered feature (DescriptorSpec), applied across the whole assembled descriptor block.

Methods

Name Description
get_descriptor_names The descriptor column names this generator produces.
get_descriptor_values Descriptor values for a series of structures (columns = descriptor names).
get_descriptor_names
data_models.descriptor_generators.api.DescriptorGenerator.get_descriptor_names()

The descriptor column names this generator produces.

get_descriptor_values
data_models.descriptor_generators.api.DescriptorGenerator.get_descriptor_values(
    values,
)

Descriptor values for a series of structures (columns = descriptor names).

Fingerprints

data_models.descriptor_generators.api.Fingerprints()

Fragments

data_models.descriptor_generators.api.Fragments()

Methods

Name Description
validate_fragments Validates that fragments have unique names
validate_fragments
data_models.descriptor_generators.api.Fragments.validate_fragments(fragments)

Validates that fragments have unique names

Parameters
Name Type Description Default
categories List[str] List of fragment names required
Raises
Name Type Description
ValueError when fragments have non-unique names
Returns
Name Type Description
List[str]: List of the fragments

MordredDescriptors

data_models.descriptor_generators.api.MordredDescriptors()

Methods

Name Description
validate_descriptors Validates that descriptors have unique names
validate_descriptors
data_models.descriptor_generators.api.MordredDescriptors.validate_descriptors(
    descriptors,
)

Validates that descriptors have unique names

Parameters
Name Type Description Default
descriptors List[str] List of descriptor names required
Raises
Name Type Description
ValueError when descriptors have non-unique names
Returns
Name Type Description
List[str]: List of the descriptors