Featurize

Class to convert SMILES to Morgan fingerprints

FeaturizeResult(items, skipped=dict())

Dataclass holding the results of a featurization pass.

items

Dictionary of compound id as key and their Morgan fingerprint as value

n_compounds

Total compounds seen (featurized + skipped).

skipped = field(default_factory=dict)

Dictionary of compound id as key and the reason for skipping as value.

MorganFeaturizer(radius=2, fp_size=2048)

Class to convert SMILES to Morgan fingerprints.

PARAMETER	DESCRIPTION
radius	Morgan radius to use . Default is 2 which is ECFP4. TYPE: int DEFAULT: 2
fp_size	Fingerprint length in bits. TYPE: int DEFAULT: 2048

fp_size = fp_size

radius = radius

featurize(raw)

Featurize a batch of SMILES keyed by their compound id.

PARAMETER	DESCRIPTION
raw	dictionary of compound id as key and their SMILES as value. TYPE: dict[str, str]

RETURNS	DESCRIPTION
FeaturizeResult	A FeaturizeResult object containing the fingerprints and any skipped compounds.

resolve_smiles(variants, *, standardize=True)

Resolve a set of SMILES mapping to the same InChIKey to a single deterministic representative SMILES string.

PARAMETER	DESCRIPTION
variants	A set of SMILES strings that map to the same InChIKey. TYPE: set[str]
standardize	Whether to standardize the molecules before comparison. Default is True. TYPE: bool DEFAULT: True

RETURNS

DESCRIPTION

tuple[str, str]

A tuple containing the chosen SMILES string and a method string indicating how the choice was made one of the following: - "single" Only one SMILES string is available for these variants. - "same_structure" variants that were parsed canonicalize to one structure so any of them can be returned. Thus, the minimum lexicographically ordered canonical SMILES of the variants is returned. - "standardized" when standardize is True, when compared in their neutral parent forms, the neutral parents agree. Thus, the canonical SMILES of the neutral parent is returned. - "standardized_picked" when standardize is True, but the neutral parents still differ. Thus, the minimum lexicographically ordered canonical SMILES of the neutral parents is returned - "picked_raw" when standardize is False or nothing standardized, the minimum lexicographically ordered canonical SMILES of the raw variants is returned.