gt4sd.algorithms.conditional_generation.paccmann_rl.core module¶

PaccMann^RL Algorithm.

PaccMann^RL generation is conditioned via reinforcement learning.

Summary¶

Classes:

`PaccMannRL`	PaccMann^RL Algorithm.
`PaccMannRLOmicBasedGenerator`	Configuration to generate compounds with low IC50 for a target omics profile.
`PaccMannRLProteinBasedGenerator`	Configuration to generate compounds with high affinity to a target protein.

Reference¶

class PaccMannRL(configuration, target)[source]¶

Bases: GeneratorAlgorithm[S, T]

PaccMann^RL Algorithm.

__init__(configuration, target)[source]¶

Instantiate PaccMannRL ready to generate items.

Parameters

configuration (AlgorithmConfiguration[~S, ~T]) – domain and application specification defining parameters, types and validations.
target (Optional[~T, None]) – a target for which to generate items.

Example

An example for generating small molecules (SMILES) with high affinity for a target protein:

affinity_config = PaccMannRLProteinBasedGenerator()
target = "MVLSPADKTNVKAAWGKVGAHAGEYGAEALERMFLSFPTT"
paccmann_affinity = PaccMannRL(configuration=affinity_config, target=target)
items = list(paccmann_affinity.sample(10))
print(items)

get_generator(configuration, target)[source]¶

Get the function to sample batches via PaccMannRL’s ConditionalGenerator.

Parameters

configuration (AlgorithmConfiguration[~S, ~T]) – helps to set up specific application of PaccMannRL.
target (Optional[~T, None]) – context or condition for the generation.

Return type

Callable[[~T], Iterable[Any]]

Returns

callable with target generating a batch of items.

validate_configuration(configuration)[source]¶

Overload to validate the a configuration for the algorithm.

Parameters: configuration (AlgorithmConfiguration[~S, ~T]) – the algorithm configuration.
Raises: InvalidAlgorithmConfiguration – in case the configuration for the algorithm is invalid.
Return type: AlgorithmConfiguration[~S, ~T]
Returns: the validated configuration.

__abstractmethods__ = frozenset({})¶

__annotations__ = {'generate': 'Untargeted', 'generator': 'Union[Untargeted, Targeted[T]]', 'max_runtime': 'int', 'max_samples': 'int', 'target': 'Optional[T]'}¶

__doc__ = 'PaccMann\\ :superscript:`RL` Algorithm.'¶

__module__ = 'gt4sd.algorithms.conditional_generation.paccmann_rl.core'¶

__orig_bases__ = (gt4sd.algorithms.core.GeneratorAlgorithm[~S, ~T],)¶

__parameters__ = (~S, ~T)¶

_abc_impl = <_abc._abc_data object>¶

class PaccMannRLProteinBasedGenerator(*args, **kwargs)[source]¶

Bases: PaccMannRLProteinBasedGenerator, Generic[T]

Configuration to generate compounds with high affinity to a target protein.

Implementation from the paper: https://doi.org/10.1088/2632-2153/abe808.

algorithm_type: ClassVar[str] = 'conditional_generation'¶: General type of generative algorithm.

domain: ClassVar[str] = 'materials'¶: General application domain. Hints at input/output types.

algorithm_version: str = 'v0'¶

To differentiate between different versions of an application.

There is no imposed naming convention.

batch_size: int = 32¶

temperature: float = 1.4¶

generated_length: int = 100¶

get_target_description()[source]¶

Get description of the target for generation.

Return type: Dict[str, str]
Returns: target description.

get_conditional_generator(resources_path)[source]¶

Instantiate the actual generator implementation.

Parameters: resources_path (str) – local path to model files.
Return type: ProteinSequenceConditionalGenerator
Returns: instance with generate_batch method for targeted generation.

validate_item(item)[source]¶

Check that item is a valid SMILES.

Parameters: item (str) – a generated item that is possibly not valid.
Raises: InvalidItem – in case the item can not be validated.
Return type: str
Returns: the validated SMILES.

__annotations__ = {'algorithm_application': 'ClassVar[str]', 'algorithm_name': 'ClassVar[str]', 'algorithm_type': typing.ClassVar[str], 'algorithm_version': <class 'str'>, 'batch_size': <class 'int'>, 'domain': typing.ClassVar[str], 'generated_length': <class 'int'>, 'temperature': <class 'float'>}¶

__dataclass_fields__ = {'algorithm_application': Field(name='algorithm_application',type=typing.ClassVar[str],default='PaccMannRLProteinBasedGenerator',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_name': Field(name='algorithm_name',type=typing.ClassVar[str],default='PaccMannRL',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_type': Field(name='algorithm_type',type=typing.ClassVar[str],default='conditional_generation',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'algorithm_version': Field(name='algorithm_version',type=<class 'str'>,default='v0',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=False,_field_type=_FIELD), 'batch_size': Field(name='batch_size',type=<class 'int'>,default=32,default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({'description': 'Batch size used for the generative model sampling.'}),kw_only=False,_field_type=_FIELD), 'domain': Field(name='domain',type=typing.ClassVar[str],default='materials',default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({}),kw_only=<dataclasses._MISSING_TYPE object>,_field_type=_FIELD_CLASSVAR), 'generated_length': Field(name='generated_length',type=<class 'int'>,default=100,default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({'description': 'Maximum length in tokens of the generated molcules (relates to the SMILES length).'}),kw_only=False,_field_type=_FIELD), 'temperature': Field(name='temperature',type=<class 'float'>,default=1.4,default_factory=<dataclasses._MISSING_TYPE object>,init=True,repr=True,hash=None,compare=True,metadata=mappingproxy({'description': 'Temperature parameter for the softmax sampling in decoding.'}),kw_only=False,_field_type=_FIELD)}¶

__dataclass_params__ = _DataclassParams(init=True,repr=True,eq=True,order=False,unsafe_hash=False,frozen=False)¶

__doc__ = '\n Configuration to generate compounds with high affinity to a target protein.\n\n Implementation from the paper: https://doi.org/10.1088/2632-2153/abe808.\n '¶

__eq__(other)¶: Return self==value.

__hash__ = None¶

__init__(*args, **kwargs)¶

__is_pydantic_dataclass__ = True¶

__match_args__ = ('algorithm_version', 'batch_size', 'temperature', 'generated_length')¶

__module__ = 'gt4sd.algorithms.conditional_generation.paccmann_rl.core'¶

__orig_bases__ = (<class 'types.PaccMannRLProteinBasedGenerator'>, typing.Generic[~T])¶

__parameters__ = (~T,)¶

__pydantic_complete__ = True¶

__pydantic_config__ = {}¶

__pydantic_core_schema__ = {'cls': <class 'gt4sd.algorithms.conditional_generation.paccmann_rl.core.PaccMannRLProteinBasedGenerator'>, 'config': {'title': 'PaccMannRLProteinBasedGenerator'}, 'fields': ['algorithm_version', 'batch_size', 'temperature', 'generated_length'], 'frozen': False, 'post_init': False, 'ref': 'types.PaccMannRLProteinBasedGenerator:94818324615808', 'schema': {'collect_init_only': False, 'computed_fields': [], 'dataclass_name': 'PaccMannRLProteinBasedGenerator', 'fields': [{'type': 'dataclass-field', 'name': 'algorithm_version', 'schema': {'type': 'default', 'schema': {'type': 'str'}, 'default': 'v0'}, 'kw_only': False, 'init': True, 'metadata': {}}, {'type': 'dataclass-field', 'name': 'batch_size', 'schema': {'type': 'default', 'schema': {'type': 'int'}, 'default': 32}, 'kw_only': False, 'init': True, 'metadata': {'pydantic_js_updates': {'description': 'Batch size used for the generative model sampling.'}}}, {'type': 'dataclass-field', 'name': 'temperature', 'schema': {'type': 'default', 'schema': {'type': 'float'}, 'default': 1.4}, 'kw_only': False, 'init': True, 'metadata': {'pydantic_js_updates': {'description': 'Temperature parameter for the softmax sampling in decoding.'}}}, {'type': 'dataclass-field', 'name': 'generated_length', 'schema': {'type': 'default', 'schema': {'type': 'int'}, 'default': 100}, 'kw_only': False, 'init': True, 'metadata': {'pydantic_js_updates': {'description': 'Maximum length in tokens of the generated molcules (relates to the SMILES length).'}}}], 'type': 'dataclass-args'}, 'slots': True, 'type': 'dataclass'}¶

__pydantic_decorators__ = DecoratorInfos(validators={}, field_validators={}, root_validators={}, field_serializers={}, model_serializers={}, model_validators={}, computed_fields={})¶

__pydantic_fields__ = {'algorithm_version': FieldInfo(annotation=str, required=False, default='v0', init=True, kw_only=False), 'batch_size': FieldInfo(annotation=int, required=False, default=32, description='Batch size used for the generative model sampling.', init=True, kw_only=False), 'generated_length': FieldInfo(annotation=int, required=False, default=100, description='Maximum length in tokens of the generated molcules (relates to the SMILES length).', init=True, kw_only=False), 'temperature': FieldInfo(annotation=float, required=False, default=1.4, description='Temperature parameter for the softmax sampling in decoding.', init=True, kw_only=False)}¶

classmethod __pydantic_fields_complete__()¶

Return whether the fields were successfully collected (i.e. type hints were successfully resolved).

This is a private helper, not meant to be used outside Pydantic.

Return type: bool

__pydantic_serializer__ = SchemaSerializer(serializer=PolymorphismTrampoline( PolymorphismTrampoline { class: Py( 0x0000563c9c8e9280, ), serializer: PolymorphismTrampoline( PolymorphismTrampoline { class: Py( 0x0000563c9c8e9280, ), serializer: Dataclass( DataclassSerializer { class: Py( 0x0000563c9c8e9280, ), serializer: Fields( GeneralFieldsSerializer { fields: { "temperature": SerField { key: "temperature", alias: None, serializer: Some( WithDefault( WithDefaultSerializer { default: Default( Py( 0x00007f214cdd1650, ), ), serializer: Float( FloatSerializer { inf_nan_mode: Null, }, ), }, ), ), required: true, serialize_by_alias: None, serialization_exclude_if: None, }, "algorithm_version": SerField { key: "algorithm_version", alias: None, serializer: Some( WithDefault( WithDefaultSerializer { default: Default( Py( 0x00007f222afe89f0, ), ), serializer: Str( StrSerializer, ), }, ), ), required: true, serialize_by_alias: None, serialization_exclude_if: None, }, "generated_length": SerField { key: "generated_length", alias: None, serializer: Some( WithDefault( WithDefaultSerializer { default: Default( Py( 0x00007f222f300d50, ), ), serializer: Int( IntSerializer, ), }, ), ), required: true, serialize_by_alias: None, serialization_exclude_if: None, }, "batch_size": SerField { key: "batch_size", alias: None, serializer: Some( WithDefault( WithDefaultSerializer { default: Default( Py( 0x00007f222f3004d0, ), ), serializer: Int( IntSerializer, ), }, ), ), required: true, serialize_by_alias: None, serialization_exclude_if: None, }, }, computed_fields: Some( ComputedFields( [], ), ), mode: SimpleDict, extra_serializer: None, filter: SchemaFilter { include: None, exclude: None, }, required_fields: 4, }, ), fields: [ Py( 0x00007f222affaf10, ), Py( 0x00007f222afe83f0, ), Py( 0x00007f21b65bb8b0, ), Py( 0x00007f214cb7e970, ), ], name: "PaccMannRLProteinBasedGenerator", }, ), enabled_from_config: false, }, ), enabled_from_config: false, }, ), definitions=[])¶

__pydantic_validator__ = SchemaValidator(title="PaccMannRLProteinBasedGenerator", validator=Dataclass( DataclassValidator { strict: false, validator: DataclassArgs( DataclassArgsValidator { fields: [ Field { kw_only: false, name: "algorithm_version", init: true, init_only: false, lookup_path_collection: LookupPathCollection { by_name: LookupPath { first_item: PathItemString( "algorithm_version", ), rest: [], }, by_alias: [], }, validator: WithDefault( WithDefaultValidator { default: Default( Py( 0x00007f222afe89f0, ), ), on_error: Raise, validator: Str( StrValidator { strict: false, coerce_numbers_to_str: false, }, ), validate_default: false, copy_default: false, name: "default[str]", undefined: Py( 0x00007f222cf13af0, ), }, ), frozen: false, }, Field { kw_only: false, name: "batch_size", init: true, init_only: false, lookup_path_collection: LookupPathCollection { by_name: LookupPath { first_item: PathItemString( "batch_size", ), rest: [], }, by_alias: [], }, validator: WithDefault( WithDefaultValidator { default: Default( Py( 0x00007f222f3004d0, ), ), on_error: Raise, validator: Int( IntValidator { strict: false, }, ), validate_default: false, copy_default: false, name: "default[int]", undefined: Py( 0x00007f222cf13af0, ), }, ), frozen: false, }, Field { kw_only: false, name: "temperature", init: true, init_only: false, lookup_path_collection: LookupPathCollection { by_name: LookupPath { first_item: PathItemString( "temperature", ), rest: [], }, by_alias: [], }, validator: WithDefault( WithDefaultValidator { default: Default( Py( 0x00007f214cdd1650, ), ), on_error: Raise, validator: Float( FloatValidator { strict: false, allow_inf_nan: true, }, ), validate_default: false, copy_default: false, name: "default[float]", undefined: Py( 0x00007f222cf13af0, ), }, ), frozen: false, }, Field { kw_only: false, name: "generated_length", init: true, init_only: false, lookup_path_collection: LookupPathCollection { by_name: LookupPath { first_item: PathItemString( "generated_length", ), rest: [], }, by_alias: [], }, validator: WithDefault( WithDefaultValidator { default: Default( Py( 0x00007f222f300d50, ), ), on_error: Raise, validator: Int( IntValidator { strict: false, }, ), validate_default: false, copy_default: false, name: "default[int]", undefined: Py( 0x00007f222cf13af0, ), }, ), frozen: false, }, ], positional_count: 4, init_only_count: None, dataclass_name: "PaccMannRLProteinBasedGenerator", validator_name: "dataclass-args[PaccMannRLProteinBasedGenerator]", extra_behavior: Ignore, extras_validator: None, loc_by_alias: true, validate_by_alias: None, validate_by_name: None, }, ), class: Py( 0x0000563c9c8e9280, ), generic_origin: None, fields: [ Py( 0x00007f222affaf10, ), Py( 0x00007f222afe83f0, ), Py( 0x00007f21b65bb8b0, ), Py( 0x00007f214cb7e970, ), ], post_init: None, revalidate: Never, name: "PaccMannRLProteinBasedGenerator", frozen: false, slots: true, }, ), definitions=[], cache_strings=True)¶

__repr__()¶: Return repr(self).

__signature__ = <Signature (algorithm_version: str = 'v0', batch_size: int = 32, temperature: float = 1.4, generated_length: int = 100) -> None>¶

__wrapped__¶: alias of PaccMannRLProteinBasedGenerator

class PaccMannRLOmicBasedGenerator(*args, **kwargs)[source]¶

Bases: PaccMannRLOmicBasedGenerator, Generic[T]

Configuration to generate compounds with low IC50 for a target omics profile.

Implementation from the paper: https://doi.org/10.1016/j.isci.2021.102269.

algorithm_type: ClassVar[str] = 'conditional_generation'¶: General type of generative algorithm.

domain: ClassVar[str] = 'materials'¶: General application domain. Hints at input/output types.