def main():
# Get file name and read the sequence
test_file, sa_file = parse_args()
test_sequence = utils.read_sequence(test_file)
sa_sequence = utils.read_sequence(sa_file)
# Read in the model
json_model = utils.read_json()
# Classify the test sequence
predicted_rsa = DecisionTree.classify_sequence(test_sequence, json_model)
utils.print_alignment(test_sequence, predicted_rsa, sa_sequence)
def needleman_wunsch(x, iterative_method, output_dir, proteins_dir):
_, protein1, protein2 = x
protein1_header, protein1_seq = read_sequence(protein1, proteins_dir)
protein2_header, protein2_seq = read_sequence(protein2, proteins_dir)
output_name = [protein1_header[1:5], protein2_header[1:5], 'global']
if iterative_method:
output_name.append('iterative')
protein1_align, protein2_align, similarity = _needleman_wunsch_iterative(protein1_seq, protein2_seq)
else:
protein1_align, protein2_align, similarity = _needleman_wunsch(protein1_seq, protein2_seq)
output = os.path.join(output_dir, '_'.join(output_name))
save_alignment(protein1_align, protein2_align, similarity, output)
def smith_waterman(x, iterative_method, output_dir, proteins_dir):
_, protein1, protein2 = x
protein1_header, protein1_seq = read_sequence(protein1, proteins_dir)
protein2_header, protein2_seq = read_sequence(protein2, proteins_dir)
output_name = [protein1_header[1:5], protein2_header[1:5], 'local']
if iterative_method:
output_name.append('iterative')
protein1_align, protein2_align, similarity = _smith_waterman_iterative(protein1_seq, protein2_seq)
else:
protein1_align, protein2_align, similarity = _smith_waterman(protein1_seq, protein2_seq)
output = os.path.join(output_dir, '_'.join(output_name))
save_alignment(protein1_align, protein2_align, similarity, output)
def parse_args():
if len(sys.argv) < 3:
print(err_msg)
sys.exit()
try:
sequence_1 = utils.read_sequence(sys.argv[1])
sequence_2 = utils.read_sequence(sys.argv[2])
except:
# File parsing has failed. Oops.
print(err_msg)
sys.exit()
return sequence_1, sequence_2
def build_feature_matrix(self):
# For each fasta file in the training data, read the sequences and add them to the feature matrix
for fasta_name in self.fasta_train:
fasta = utils.read_sequence(fasta_name, self.fasta_dir)
sa = utils.read_sequence(fasta_name.replace('.fasta', '.sa'),
self.sa_dir)
for index in range(len(fasta)):
# Create the AA object
acid = get_amino_acid(fasta[index].upper())
# Add the RSA label
rsa_binary = rsa_labels[sa[index]]
acid['rsa-label'] = rsa_binary
# Add the acid to the matrix
self.feature_matrix.append(acid)
def evaluate_model(self):
# Keeping track of metrics (true condition, predicted condition)
metrics = {
(0, 0): 0, # True negative
(0, 1): 0, # False positive
(1, 0): 0, # False negative
(1, 1): 0 # True positive
}
# For each fasta file in the testing data, walk the tree for each amino acid
for fasta_name in self.fasta_test:
fasta = utils.read_sequence(fasta_name, self.fasta_dir)
sa = utils.read_sequence(fasta_name.replace('.fasta', '.sa'),
self.sa_dir)
for index in range(len(fasta)):
# Test this amino acid against our decision tree
amino_acid = fasta[index]
expected_result = rsa_labels[sa[index]]
calculated_result = self.walk_tree(get_amino_acid(amino_acid))
# print('Acid {}, expected {}, calculated {}'.format(amino_acid, expected_result, calculated_result))
metrics[expected_result, calculated_result] += 1
self.calculate_eval_metrics(metrics)