schrodinger.seam.yaml.decorators module

schrodinger.seam.yaml.decorators.yaml_transform(cls: Optional[Type[PTransform]] = None, *, name: Optional[str] = None) → Union[Type[PTransform], Callable[[Type[PTransform]], Type[PTransform]]]

Decorator to register a PTransform for use in SEAM YAML pipelines.

Decorated transforms are automatically discovered by the SEAM YAML provider system and made available for use in YAML pipeline definitions.

Example usage:

@yaml_transform(name="LigFilter")
class FilterLigands(PTransformWithConfig):
    ...

@yaml_transform  # Uses class name as YAML name
class MyTransform(beam.PTransform):
    ...

Parameters:

• cls – The transform class (when used without parentheses).

• name – Optional custom name for the transform in YAML. If not provided, the class name is used.

Returns:

The decorated class (unchanged, but registered in the YAML registry).

schrodinger.seam.yaml.decorators.get_yaml_transforms() → Dict[str, Type[PTransform]]

Get all registered YAML transforms.

This function is called by the SEAM YAML provider system to discover transforms that have been decorated with @yaml_transform.

Returns:

Dictionary mapping YAML names to transform classes.

class schrodinger.seam.yaml.decorators.SchemadStructureWrapper(ptransform_instance, input_field='structure', output_field='structure')

Bases: PTransform

Wrapper PTransform for applying Structure-based transforms to Beam Rows.

This is used by legacy Row-based transforms that need to extract a Structure from a Row field, process it, and put it back into a Row.

For new transforms, prefer working directly with Structure objects using the @yaml_transform decorator.

ROW_METADATA_FIELD = 's_seam_row_metadata'

__init__(ptransform_instance, input_field='structure', output_field='structure')

expand(pcoll: PCollection)