sample

sample#

Factory for creating sequence sampling and optimization functions.

prxteinmpnn.sampling.sample.make_sample_sequences(model_parameters, decoding_order_fn, sampling_strategy='temperature', num_encoder_layers=3, num_decoder_layers=3)[source]#

Create a function to sample or optimize sequences from a structure.

Return type:

Callable[..., tuple[Int[Array, 'num_residues'], Float[Array, 'num_residues num_classes'], Int[Array, 'num_residues']]]

Parameters:

  • model_parameters (PyTree[str, 'P'])

  • decoding_order_fn (Callable[[Unpack[tuple[Key[Array, ''] | UInt32[Array, '2'], Int]]], tuple[Int[Array, 'num_residues'], Key[Array, ''] | UInt32[Array, '2']]])

  • sampling_strategy (Literal['temperature', 'straight_through'])

  • num_encoder_layers (int)

  • num_decoder_layers (int)

Contents