Auto3D.config.Auto3DOptions

class Auto3D.config.Auto3DOptions(path: str | None = None, k: int | bool = False, window: float | bool = False, verbose: bool = False, job_name: str = '', enumerate_tautomer: bool = False, tauto_engine: str = 'rdkit', pKaNorm: bool = True, isomer_engine: str = 'rdkit', enumerate_isomer: bool = True, mode_oe: str = 'classic', mpi_np: int = 4, max_confs: int | None = None, use_gpu: bool = True, gpu_idx: int | list[int] = 0, capacity: int = 42, optimizing_engine: str = 'AIMNET', patience: int = 250, opt_steps: int = 2000, convergence_threshold: float = 0.01, threshold: float = 0.3, memory: int | None = None, batchsize_atoms: int = 1024, allow_tf32: bool = False)

Bases: object

Configuration options for Auto3D conformer generation.

This dataclass provides proper type hints, validation, and IDE support for Auto3D configuration.

Example

>>> from Auto3D import Auto3DOptions, main
>>> config = Auto3DOptions(path="input.smi", k=5)
>>> result = main(config)

__init__(path: str | None = None, k: int | bool = False, window: float | bool = False, verbose: bool = False, job_name: str = '', enumerate_tautomer: bool = False, tauto_engine: str = 'rdkit', pKaNorm: bool = True, isomer_engine: str = 'rdkit', enumerate_isomer: bool = True, mode_oe: str = 'classic', mpi_np: int = 4, max_confs: int | None = None, use_gpu: bool = True, gpu_idx: int | list[int] = 0, capacity: int = 42, optimizing_engine: str = 'AIMNET', patience: int = 250, opt_steps: int = 2000, convergence_threshold: float = 0.01, threshold: float = 0.3, memory: int | None = None, batchsize_atoms: int = 1024, allow_tf32: bool = False) → None

Methods

__init__([path, k, window, verbose, ...])
get(key[, default])	Dict-like get method for backward compatibility.
items()	Return field name-value pairs for backward compatibility.
keys()	Return field names for backward compatibility.
to_optimization_config()	Create an OptimizationConfig from these options.

Attributes

allow_tf32	Enable TF32 for faster matmul on Ampere+ GPUs (less precise).
batchsize_atoms	Number of atoms per optimization batch per GB.
capacity	Number of SMILES handled per 1GB memory.
convergence_threshold	Optimization converges when max force is below this (eV/Å).
enumerate_isomer	When True, enumerate cis/trans and R/S isomers.
enumerate_tautomer	When True, enumerate tautomers for the input.
gpu_idx	GPU device index or list of indices.
isomer_engine	'rdkit' or 'omega'.
job_name	Folder name to save all metadata.
k	Output top-k structures for each SMILES.
max_confs	Maximum conformers per SMILES.
memory	RAM size assigned to Auto3D in GB.
mode_oe	'classic', 'macrocycle', 'dense', 'pose', 'rocs', or 'fast_rocs'.
mpi_np	Number of CPU cores for isomer generation.
opt_steps	Maximum optimization steps per structure.
optimizing_engine	'ANI2x', 'ANI2xt', 'AIMNET', or path to custom NNP.
pKaNorm	Normalize ionization state to pH ~7.4 (only with tauto_engine='oechem').
path	Path to input .smi or .sdf file containing SMILES/molecules.
patience	Drop conformer from optimization if force doesn't decrease for this many steps.
tauto_engine	'rdkit' or 'oechem'.
threshold	RMSD threshold for duplicate removal (Å).
use_gpu	Use GPU when available.
verbose	When True, save all metadata while running.
window	Output structures within x kcal/mol of lowest energy.

path: str | None = None

Path to input .smi or .sdf file containing SMILES/molecules.

k: int | bool = False

Output top-k structures for each SMILES. Set to int or False.

window: float | bool = False

Output structures within x kcal/mol of lowest energy. Set to float or False.

verbose: bool = False

When True, save all metadata while running.

job_name: str = ''

Folder name to save all metadata.

enumerate_tautomer: bool = False

When True, enumerate tautomers for the input.

tauto_engine: str = 'rdkit'

‘rdkit’ or ‘oechem’.

Type:

Program to enumerate tautomers

pKaNorm: bool = True

Normalize ionization state to pH ~7.4 (only with tauto_engine=’oechem’).

isomer_engine: str = 'rdkit'

‘rdkit’ or ‘omega’.

Type:

Program for generating 3D isomers

enumerate_isomer: bool = True

When True, enumerate cis/trans and R/S isomers.

mode_oe: str = 'classic'

‘classic’, ‘macrocycle’, ‘dense’, ‘pose’, ‘rocs’, or ‘fast_rocs’.

Type:

Omega mode

mpi_np: int = 4

Number of CPU cores for isomer generation.

max_confs: int | None = None

Maximum conformers per SMILES. None uses dynamic number (num_heavy_atoms - 1).

use_gpu: bool = True

Use GPU when available.

gpu_idx: int | list[int] = 0

GPU device index or list of indices.

capacity: int = 42

Number of SMILES handled per 1GB memory.

optimizing_engine: str = 'AIMNET'

‘ANI2x’, ‘ANI2xt’, ‘AIMNET’, or path to custom NNP.

Type:

Model for optimization

patience: int = 250

Drop conformer from optimization if force doesn’t decrease for this many steps.

opt_steps: int = 2000

Maximum optimization steps per structure.

convergence_threshold: float = 0.01

Optimization converges when max force is below this (eV/Å).

threshold: float = 0.3

RMSD threshold for duplicate removal (Å).

memory: int | None = None

RAM size assigned to Auto3D in GB. None for automatic detection.

batchsize_atoms: int = 1024

Number of atoms per optimization batch per GB.

allow_tf32: bool = False

Enable TF32 for faster matmul on Ampere+ GPUs (less precise). Default False.

get(key: str, default=None)

Dict-like get method for backward compatibility.

keys()

Return field names for backward compatibility.

items()

Return field name-value pairs for backward compatibility.

to_optimization_config() → OptimizationConfig

Create an OptimizationConfig from these options.

Returns:

OptimizationConfig with values from this Auto3DOptions instance.