openfold.data.feature_pipeline

Classes

FeaturePipeline(config)

Functions

make_data_config(config, mode, num_res)
np_example_to_features(np_example, config, mode)
np_to_tensor_dict(np_example, features)	Creates dict of tensors from a dict of NumPy arrays.

class FeaturePipeline(config)

Parameters:

config (ConfigDict)

process_features(raw_features, mode='train', is_multimer=False)

Parameters:

• raw_features (Mapping[str, ndarray])

• mode (str)

• is_multimer (bool)

Return type:

Mapping[str, ndarray]

make_data_config(config, mode, num_res)

Parameters:

• config (ConfigDict)

• mode (str)

• num_res (int)

Return type:

Tuple[ConfigDict, List[str]]

np_example_to_features(np_example, config, mode, is_multimer=False)

Parameters:

• np_example (Mapping[str, ndarray])

• config (ConfigDict)

• mode (str)

• is_multimer (bool)

np_to_tensor_dict(np_example, features)

Creates dict of tensors from a dict of NumPy arrays.

Parameters:

• np_example (Mapping[str, ndarray]) – A dict of NumPy feature arrays.

• features (Sequence[str]) – A list of strings of feature names to be returned in the dataset.

Returns:

A dictionary of features mapping feature names to features. Only the given features are returned, all other ones are filtered out.

Return type:

Dict[str, Tensor]