def test_create_feature_vectors(self):
create_positive_and_negative_dataset(1, 10)
pssms_file = "/tmp/pssms.txt"
conservation = True
pssmData = feature.parse_pssms_file(pssms_file)
pssm = pssmData.get_PSSMRecord(pssmData.get_uniprotURIs()[0])
feature_vectors = feature.create_feature_vectors(pssm, 1, conservation=True)
self.assertEqual(66, len(feature_vectors[0]))
self.assertEqual(0, feature_vectors[0][-3])
self.assertEqual(0, feature_vectors[9][-1])
def write_to_output_file(prediction_output_file, user_defined_header, results):
with open(prediction_output_file, 'a') as fp:
idx = 0
for predicted_label, decision_value in zip(results['label'],
results['decision_value']):
fp.write("{} {} {} {}\n".format(user_defined_header, idx,
predicted_label, decision_value))
idx += 1
if __name__ == "__main__":
arguments = docopt(__doc__)
best_chromosome_file = arguments['<best_chromosome_file>']
pssms_file = arguments['<pssms_file>']
pickled_model_file = arguments['<pickled_model_file>']
prediction_output_file = arguments['<prediction_output_file>']
method_genes_decision_value = common.get_method_genes_decision_value(
best_chromosome_file)
method, window_size, my_decision_value = method_genes_decision_value[
0], int(method_genes_decision_value[3]), float(
method_genes_decision_value[4])
pssmData = feature.parse_pssms_file(pssms_file)
for user_defined_header in pssmData.get_uniprotURIs():
pssm = pssmData.get_PSSMRecord(user_defined_header)
test_dataset = feature.create_feature_vectors(pssm, window_size)
results = predict(method, pickled_model_file, my_decision_value,
test_dataset)
write_to_output_file(prediction_output_file, user_defined_header,
results)
return predict_with_NN_classifier(clf_or_net, my_decision_value, test_dataset)
elif method == "randomForest":
return predict_with_RF_classifier(clf_or_net, my_decision_value, test_dataset)
elif method == "SVM":
return predict_with_SVM_classifier(clf_or_net, my_decision_value, test_dataset)
else:
raise ValueError("method must be neuralNetwork or randomForest or SVM [{}]".format(method))
def write_to_output_file(prediction_output_file, user_defined_header, results):
with open(prediction_output_file, 'a') as fp:
idx = 0
for predicted_label, decision_value in zip(results['label'], results['decision_value']):
fp.write("{} {} {} {}\n".format(user_defined_header, idx, predicted_label, decision_value))
idx += 1
if __name__ == "__main__":
arguments = docopt(__doc__)
best_chromosome_file = arguments['<best_chromosome_file>']
pssms_file = arguments['<pssms_file>']
pickled_model_file = arguments['<pickled_model_file>']
prediction_output_file = arguments['<prediction_output_file>']
method_genes_decision_value = common.get_method_genes_decision_value(best_chromosome_file)
method, window_size, my_decision_value = method_genes_decision_value[0], int(method_genes_decision_value[3]), float(method_genes_decision_value[4])
pssmData = feature.parse_pssms_file(pssms_file)
for user_defined_header in pssmData.get_uniprotURIs():
pssm = pssmData.get_PSSMRecord(user_defined_header)
test_dataset = feature.create_feature_vectors(pssm, window_size)
results = predict(method, pickled_model_file, my_decision_value, test_dataset)
write_to_output_file(prediction_output_file, user_defined_header, results)
