main

main#

Functions

fix_seed(seed)

Seed all necessary random number generators.

log_print(text[, color, on_color, attrs])

Print a message to the console with optional color formatting.

main(args)

Main entry point for training the BRIDGE model.
main.log_print(text, color=None, on_color=None, attrs=None)[source]#

Print a message to the console with optional color formatting.

This utility function attempts to use third-party libraries (termcolor and pycrayon) to produce colored or styled console output. If these libraries are not available, it gracefully falls back to standard print without formatting.

Parameters:

  • text (str) – The message to be printed to the console.

  • color (str, optional) – Text color name supported by termcolor (e.g., ‘red’, ‘green’). If None, the default terminal color is used.

  • on_color (str, optional) – Background color name supported by termcolor (e.g., ‘on_blue’, ‘on_yellow’).

  • attrs (list of str, optional) – List of text attributes supported by termcolor, such as [‘bold’, ‘underline’].

main.fix_seed(seed)[source]#

Seed all necessary random number generators.

main.main(args)[source]#

Main entry point for training the BRIDGE model.

This function orchestrates the full training pipeline, including:

  • random seed initialization

  • device (CPU/GPU) configuration

  • data loading and preprocessing

  • sequence embedding extraction using a pretrained transformer

  • construction of structural, biochemical, and motif prior features

  • dataset splitting and DataLoader creation

  • model training, validation, learning-rate scheduling, and early stopping

  • model checkpointing and performance reporting

  • persistent logging/config/metrics under results/{logs,model,metrics}

Parameters:

args (argparse.Namespace) – Parsed command-line arguments specifying runtime configuration. Expected attributes include (but are not limited to):

  • seedint

    Random seed for reproducibility.

  • use_cpubool

    Whether to force CPU execution.

  • device_numint

    GPU device index to use when CUDA is available.

  • trainbool

    Whether to run the training procedure.

  • data_filestr

    Dataset identifier used to locate input files.

  • data_pathstr

    Root directory containing input FASTA and feature files.

  • Transformer_pathstr

    Path to the pretrained RBPformer model.

  • lrfloat

    Initial learning rate for the optimizer.

  • early_stoppingint

    Number of epochs without improvement before early stopping.

Contents