MolecularDiffusion.runmodes.data.featurization¶

Featurization module for MolCraft. Handles generating vectorial representations (Morgan fingerprints, SOAP) from 3D molecular data.

Attributes¶

`Chem`
`logger`
`save_safetensors`

Functions¶

`featurize`(input_source, output_dir[, method, format, ...])	Main featurization driver.
`generate_morgan`(→ Tuple[numpy.ndarray, List[str]])	Generates Morgan fingerprints for a list of entries.
`generate_soap`(→ Tuple[numpy.ndarray, List[str]])	Generates SOAP descriptors.
`save_features`(features, ids, output_dir[, format])	Saves features and indices.

Module Contents¶

MolecularDiffusion.runmodes.data.featurization.featurize(input_source: str, output_dir: str, method: str = 'morgan', format: str = 'npy', readout: str = 'mean', smilify_method: str = 'hybrid', radius: int = 2, nbits: int = 2048, rcut: float = 5.0, nmax: int = 8, lmax: int = 6)¶: Main featurization driver.

MolecularDiffusion.runmodes.data.featurization.generate_morgan(entries: List[Dict], radius: int = 2, nbits: int = 2048, smilify_method: str = 'hybrid', n_jobs: int = 1) → Tuple[numpy.ndarray, List[str]]¶: Generates Morgan fingerprints for a list of entries. Entries is a list of dicts with ‘file’ (path) or ‘data’ (atoms/coords) and ‘id’.

MolecularDiffusion.runmodes.data.featurization.generate_soap(entries: List[Dict], rcut: float = 5.0, nmax: int = 8, lmax: int = 6, readout: str = 'mean', species: List[str] = None) → Tuple[numpy.ndarray, List[str]]¶: Generates SOAP descriptors.

MolecularDiffusion.runmodes.data.featurization.save_features(features: numpy.ndarray, ids: List[str], output_dir: pathlib.Path, format: str = 'npy')¶: Saves features and indices.

MolecularDiffusion.runmodes.data.featurization.Chem = None¶

MolecularDiffusion.runmodes.data.featurization.logger¶

MolecularDiffusion.runmodes.data.featurization.save_safetensors = None¶