def test_should_return_original_alignment_size(self):
# setup
problem = MSA(score_list=[])
problem.identifiers = ['seq1', 'seq2', 'seq3']
problem.number_of_variables = 3
msa = MSASolution(problem,
msa=[('seq1', 'AC---TGAC'), ('seq2', 'AT--CT--C'),
('seq3', 'AAC---TGC')])
# check
self.assertEqual(9, msa.get_length_of_alignment())
def __find_symbol_position_in_original_sequence(self,
solution: MSASolution,
seq_index: int,
position: int):
""" Given a symbol position, finds the corresponding position of the symbol in the original
sequence if gaps are not taken into account. If the symbol is a gap the returned value is -1 """
if position > solution.get_length_of_alignment():
raise Exception(
'Position {0} is larger than the sequence size {1}'.format(
position, solution.get_length_of_alignment()))
if not solution.is_gap_char_at_sequence(seq_index, position):
symbol_position = solution.get_char_position_in_original_sequence(
seq_index, position)
else:
position = solution.get_next_char_position_after_gap(
seq_index, position)
if position < 0:
symbol_position = -1
else:
symbol_position = solution.get_char_position_in_original_sequence(
seq_index, position)
return symbol_position
def do_mutation(self, solution: MSASolution) -> MSASolution:
if random.random() <= self.probability:
length_of_alignment = solution.get_length_of_alignment()
for seq_index in range(solution.number_of_variables):
point = random.randint(0, length_of_alignment - 1)
solution.add_gap_to_sequence_at_index(seq_index, point)
if self.remove_full_of_gap_columns:
solution.remove_full_of_gaps_columns()
# Sanity check: alignment is valid (same length for all sequences)
if not solution.is_valid_msa():
raise Exception("Mutated solution is not valid! {0}".format(
solution.decode_alignment_as_list_of_pairs()))
return solution
