deep_folding.utils package¶

Submodules¶

class deep_folding.utils.pytorchtools.EarlyStopping(root_dir, patience=7, verbose=False, delta=0)¶

Bases: object

Early stops the training if validation loss doesn’t improve after a given patience.

save_checkpoint(val_loss, model, root_dir)¶: Saves model when validation loss decrease.

deep_folding.utils.save_results.create_folder(root_dir)¶: Creates folder in whiche results will be saved in : directory where to create the folder out : / creatin of a folder

deep_folding.utils.save_results.save_logs(input_dict)¶: Creates and saves a log file with all information of the run in : dico like {module:’python file’

input: {date : ‘’, data: ‘split’ or ‘all’, loss_type :’‘, batch_size: ‘’, nb_epoch:’‘, root_dir:’‘, input_type:’jeff crops or normalized crops’, skeleton: ‘True or False’}, output: {loss_train:’‘, loss_test:’‘}}

Splits the subjects IDs in a train set and a test set.

Each set is represented as a csv file (train.csv and test.csv). The first column on each row is the subject ID.

deep_folding.utils.split_train_test.get_basename_without_extension(filename)¶: Returns file basename without extension

deep_folding.utils.split_train_test.main(argv)¶

Reads argument line and creates cropped files and pickle file

Parameters:	argv – a list containing command line arguments

deep_folding.utils.split_train_test.parse_args(argv)¶

Function parsing command-line arguments

Parameters:	argv – a list containing command line arguments
Returns:	Namespace
Return type:	args

deep_folding.utils.split_train_test.split_train_test(src_dir, tgt_dir, nb_test_subjects)¶: Splits the subject IDs in train and test