# -*- coding: utf-8 -*- """ Created on Mon Jun 8 09:32:26 2020 @author: Muammer """ import os script_dir = os.path.dirname(os.path.abspath(__file__)) import numpy as np from sklearn.model_selection import cross_validate from sklearn.model_selection import cross_val_predict from sklearn.metrics import matthews_corrcoef from sklearn.metrics import classification_report from sklearn.multiclass import OneVsRestClassifier from sklearn import linear_model from sklearn.metrics import f1_score from sklearn.metrics import confusion_matrix from sklearn.model_selection import train_test_split import pandas as pd from numpy import save from sklearn.metrics import precision_recall_fscore_support from tqdm import tqdm from sklearn.metrics import accuracy_score import math representation_name = "" representation_path = "" dataset = "nc" detailed_output = False def convert_dataframe_to_multi_col(representation_dataframe): entry = pd.DataFrame(representation_dataframe['Entry']) vector = pd.DataFrame(list(representation_dataframe['Vector'])) multi_col_representation_vector = pd.merge(left=entry,right=vector,left_index=True, right_index=True) return multi_col_representation_vector def class_based_scores(c_report, c_matrix): c_report = pd.DataFrame(c_report).transpose() #print(c_report) c_report = c_report.drop(['precision', 'recall'], axis=1) c_report = c_report.drop(labels=['accuracy', 'macro avg', 'weighted avg'], axis=0) cm = c_matrix.astype('float') / c_matrix.sum(axis=1)[:, np.newaxis] #print(c_report) accuracy = cm.diagonal() #print(accuracy) #if len(accuracy) == 6: # accuracy = np.delete(accuracy, 5) accuracy = pd.Series(accuracy, index=c_report.index) c_report['accuracy'] = accuracy total = c_report['support'].sum() #print(total) num_classes = np.shape(c_matrix)[0] mcc = np.zeros(shape=(num_classes,), dtype='float32') weights = np.sum(c_matrix, axis=0)/np.sum(c_matrix) total_tp = 0 total_fp = 0 total_fn = 0 total_tn = 0 for j in range(num_classes): tp = np.sum(c_matrix[j, j]) fp = np.sum(c_matrix[j, np.concatenate((np.arange(0, j), np.arange(j+1, num_classes)))]) fn = np.sum(c_matrix[np.concatenate((np.arange(0, j), np.arange(j+1, num_classes))), j]) tn = int(total - tp - fp - fn) total_tp = total_tp + tp total_fp = total_fp + fp total_fn = total_fn + fn total_tn = total_tn + tn #print(tp,fp,fn,tn) mcc[j] = ((tp*tn)-(fp*fn))/math.sqrt((tp+fp)*(tp+fn)*(tn+fp)*(tn+fn)) #print(mcc) #if len(mcc) == 6: # mcc = np.delete(mcc, 5) mcc = pd.Series(mcc, index=c_report.index) c_report['mcc'] = mcc #c_report.to_excel('../results/resultss_class_based_'+dataset+'.xlsx') #print(c_report) return c_report, total_tp, total_fp, total_fn, total_tn def score_protein_rep(dataset): #def score_protein_rep(pkl_data_path): vecsize = 0 #protein_list = pd.read_csv('../data/auxilary_input/entry_class.csv') protein_list = pd.read_csv(os.path.join(script_dir, '../data/preprocess/entry_class_nn.csv')) dataframe = pd.read_csv(representation_path) #dataframe = convert_dataframe_to_multi_col(dataframe) #dataframe = pd.read_pickle(pkl_data_path) vecsize = dataframe.shape[1]-1 x = np.empty([0, vecsize]) xemp = np.zeros((1, vecsize), dtype=float) y = [] ne = [] print("\n\nPreprocessing data for drug-target protein family prediction...\n ") for index, row in tqdm(protein_list.iterrows(), total=len(protein_list)): pdrow = dataframe.loc[dataframe['Entry'] == row['Entry']] if len(pdrow) != 0: a = pdrow.loc[ : , pdrow.columns != 'Entry'] a = np.array(a) a.shape = (1,vecsize) x = np.append(x, a, axis=0) y.append(row['Class']) else: ne.append(index) x = np.append(x, xemp, axis=0,) y.append(0.0) #print(index) x = x.astype(np.float64) y = np.array(y) y = y.astype(np.float64) #print(len(y)) scoring = ['precision_weighted', 'recall_weighted', 'f1_weighted', 'accuracy'] target_names = ['Enzyme', 'Membrane receptor', 'Transcription factor', 'Ion channel', 'Other'] labels = [1.0, 11.0, 12.0, 1005.0, 2000.0] f1 = [] accuracy = [] mcc = [] f1_perclass = [] ac_perclass = [] mcc_perclass = [] sup_perclass = [] report_list = [] train_index = pd.read_csv(os.path.join(script_dir, '../data/preprocess/indexes/'+dataset+'_trainindex.csv')) test_index = pd.read_csv(os.path.join(script_dir, '../data/preprocess/indexes/testindex_family.csv')) train_index = train_index.dropna(axis=1) test_index = test_index.dropna(axis=1) #print(train_index) #for index in ne: conf = pd.DataFrame() print('Producing protein family predictions...\n') for i in tqdm(range(10)): clf = linear_model.SGDClassifier(class_weight="balanced", loss="log", penalty="elasticnet", max_iter=1000, tol=1e-3,random_state=i,n_jobs=-1) clf2 = OneVsRestClassifier(clf,n_jobs=-1) #print(test_index) train_indexx = train_index.iloc[i].astype(int) test_indexx = test_index.iloc[i].astype(int) #print(train_indexx) #train_indexx.drop(labels=ne) #print(type(train_indexx)) for index in ne: train_indexx = train_indexx[train_indexx!=index] test_indexx = test_indexx[test_indexx!=index] train_X, test_X = x[train_indexx], x[test_indexx] train_y, test_y = y[train_indexx], y[test_indexx] clf2.fit(train_X, train_y) #print(train_X) y_pred = clf2.predict(test_X) #y_pred = cross_val_predict(clf2, x, y, cv=10, n_jobs=-1) #mcc.append(matthews_corrcoef(test_y, y_pred, sample_weight = test_y)) f1_ = f1_score(test_y, y_pred, average='weighted') f1.append(f1_) ac = accuracy_score(test_y, y_pred) accuracy.append(ac) c_report = classification_report(test_y, y_pred, target_names=target_names, output_dict=True) c_matrix = confusion_matrix(test_y, y_pred, labels=labels) conf = conf.append(pd.DataFrame(c_matrix, columns=['Enzymes', 'Membrane receptor', 'Transcription factor', 'Ion channel', 'Other']), ignore_index=True) class_report, tp, fp, fn, tn = class_based_scores(c_report, c_matrix) #print(total_tp) mcc.append(((tp*tn)-(fp*fn))/math.sqrt((tp+fp)*(tp+fn)*(tn+fp)*(tn+fn))) f1_perclass.append(class_report['f1-score']) ac_perclass.append(class_report['accuracy']) mcc_perclass.append(class_report['mcc']) sup_perclass.append(class_report['support']) report_list.append(class_report) if detailed_output: conf.to_csv(os.path.join(script_dir, '../results/Drug_target_protein_family_classification_confusion_'+dataset+'_'+representation_name+'.csv'), index=None) f1_perclass = pd.concat(f1_perclass, axis=1) ac_perclass = pd.concat(ac_perclass, axis=1) mcc_perclass = pd.concat(mcc_perclass, axis=1) sup_perclass = pd.concat(sup_perclass, axis=1) report_list = pd.concat(report_list, axis=1) report_list.to_csv(os,path,join(script_dir, '../results/Drug_target_protein_family_classification_class_based_results_'+dataset+'_'+representation_name+'.csv')) report = pd.DataFrame() f1mean = np.mean(f1, axis=0) #print(f1mean) f1mean = f1mean.round(decimals=5) f1std = np.std(f1).round(decimals=5) acmean = np.mean(accuracy, axis=0).round(decimals=5) acstd = np.std(accuracy).round(decimals=5) mccmean = np.mean(mcc, axis=0).round(decimals=5) mccstd = np.std(mcc).round(decimals=5) labels = ['Average Score', 'Standard Deviation'] report['Protein Family'] = labels report['F1_score'] = [f1mean, f1std] report['Accuracy'] = [acmean, acstd] report['MCC'] = [mccmean, mccstd] report.to_csv(os.path.join(script_dir, '../results/Drug_target_protein_family_classification_mean_results_'+dataset+'_'+representation_name+'.csv',index=False)) #report.to_csv('scores_general.csv') #print(report) if detailed_output: save('../results/Drug_target_protein_family_classification_f1_'+dataset+'_'+representation_name+'.npy', f1) save('../results/Drug_target_protein_family_classification_accuracy_'+dataset+'_'+representation_name+'.npy', accuracy) save('../results/Drug_target_protein_family_classification_mcc_'+dataset+'_'+representation_name+'.npy', mcc) save('../results/Drug_target_protein_family_classification_class_based_f1_'+dataset+'_'+representation_name+'.npy', f1_perclass) save('../results/Drug_target_protein_family_classification_class_based_accuracy_'+dataset+'_'+representation_name+'.npy', ac_perclass) save('../results/Drug_target_protein_family_classification_class_based_mcc_'+dataset+'_'+representation_name+'.npy', mcc_perclass) save('../results/Drug_target_protein_family_classification_class_based_support_'+dataset+'_'+representation_name+'.npy', sup_perclass)