from os.path import abspath

from os.path import split

from simdeep.simdeep_boosting import SimDeepBoosting

def test_instance():

    """

    example of SimDeepBoosting

    """

    PATH_DATA = '{0}/../examples/data/'.format(split(abspath(__file__))[0])

    #Input file

    TRAINING_TSV = {'RNA': 'rna_dummy.tsv', 'METH': 'meth_dummy.tsv'}

    SURVIVAL_TSV = 'survival_dummy.tsv'

    # Optional metadata FILE

    OPTIONAL_METADATA = "metadata_dummy.tsv"

    # Subsetting training set with only males from metadata:

    SUBSET_TRAINING_WITH_META = {'stage': ['I', 'II', 'III']}

    PROJECT_NAME = 'TestProject'

    SEED = 3

    nb_it = 5 # Number of models to be built

    nb_threads = 2 # Number of processes to be used to fit individual survival models

    ################ AUTOENCODER PARAMETERS ################

    EPOCHS = 10

    ## Additional parameters for the autoencoders can be defined, see config.py file for details

    #########################################################

    ################ ADDITIONAL PARAMETERS ##################

    # PATH_TO_SAVE_MODEL = '/home/username/deepprog'

    # PVALUE_THRESHOLD = 0.01

    # NB_SELECTED_FEATURES = 10

    # STACK_MULTI_OMIC = False

    #########################################################

    # IT is possible to define a custom normalisation

    # from sklearn.preprocessing import RobustScaler

    # norm = {

    #         'CUSTOM': RobustScaler,

    # }

    boosting = SimDeepBoosting(

        nb_threads=nb_threads,

        nb_it=nb_it,

        split_n_fold=3,

        survival_tsv=SURVIVAL_TSV,

        training_tsv=TRAINING_TSV,

        # metadata_tsv=OPTIONAL_METADATA, # optional

        path_data=PATH_DATA,

        project_name=PROJECT_NAME,

        path_results=PATH_DATA,

        use_r_packages=False, # to use R functions from the survival and survcomp packages

        epochs=EPOCHS,

        seed=SEED,

        # normalization=norm,

        cluster_method='coxPH',

        metadata_usage='labels',

        use_autoencoders=True,

        feature_surv_analysis=True,

        feature_selection_usage="lasso",

        # subset_training_with_meta=SUBSET_TRAINING_WITH_META,

        # stack_multi_omic=True,

        # path_to_save_model=PATH_TO_SAVE_MODEL,

        # pvalue_threshold=PVALUE_THRESHOLD,

        # nb_selected_features=NB_SELECTED_FEATURES,

    )

    boosting.fit()

    boosting.predict_labels_on_full_dataset()

    boosting.save_models_classes()

    boosting.save_cv_models_classes()

    boosting.compute_clusters_consistency_for_full_labels()

    boosting.evalutate_cluster_performance()

    boosting.collect_cindex_for_test_fold()

    boosting.collect_cindex_for_full_dataset()

    boosting.compute_feature_scores_per_cluster()

    boosting.compute_survival_feature_scores_per_cluster(pval_thres=0.10)

    boosting.write_feature_score_per_cluster()

    boosting.collect_number_of_features_per_omic()

    boosting.compute_pvalue_for_merged_test_fold()

    boosting.load_new_test_dataset(

        tsv_dict={'RNA': 'rna_dummy.tsv'}, # OMIC file of the test set. It doesnt have to be the same as for training

        path_survival_file='survival_dummy.tsv', # Optional survival file of the test set for computing validation log-rank pvalue

        fname_key='dummy', # Name of the test test to be used

    )

    boosting.predict_labels_on_test_dataset()

    boosting.save_test_models_classes()

    boosting.compute_c_indexes_for_test_dataset()

    boosting.compute_clusters_consistency_for_test_labels()

    # Experimental method to plot the test dataset amongst the class kernel densities

    boosting.plot_supervised_kernel_for_test_sets()

    boosting.plot_supervised_predicted_labels_for_test_sets()

    boosting.load_new_test_dataset(

        tsv_dict={'METH': 'meth_dummy.tsv'}, # OMIC file of the second test set.

        path_survival_file='survival_dummy.tsv', # Survival file of the test set

        fname_key='dummy_METH', # Name of the second test test

        metadata_file="metadata_dummy.tsv" # Optional metadata

    )

    boosting.predict_labels_on_test_dataset()

    boosting.compute_c_indexes_for_test_dataset()

    boosting.compute_clusters_consistency_for_test_labels()

    # Experimental method to plot the test dataset amongst the class kernel densities

    boosting.plot_supervised_kernel_for_test_sets()

    boosting.plot_supervised_predicted_labels_for_test_sets()

if __name__ == '__main__':

    test_instance()