def fill_sequences_with_gaps_to_reach_the_max_sequence_length(
self, solution: MSASolution, max_length: int,
cutting_points: list):
for i in range(solution.number_of_variables):
sequence_length = solution.get_length_of_sequence(i)
if sequence_length != max_length:
for j in range(sequence_length, max_length):
if cutting_points[i] == -1:
solution.add_gap_to_sequence_at_index(
seq_index=i, gap_position=sequence_length - 1)
else:
solution.add_gap_to_sequence_at_index(
seq_index=i, gap_position=cutting_points[i] + 1)
def test_should_create_new_gaps_group(self):
# setup
problem = MSA(score_list=[])
problem.identifiers = ['seq1']
problem.number_of_variables = 1
msa_1 = MSASolution(problem, msa=[('seq1', 'A-')])
msa_2 = MSASolution(problem, msa=[('seq1', '-A')])
msa_3 = MSASolution(problem, msa=[('seq1', 'A-C')])
msa_4 = MSASolution(problem, msa=[('seq1', 'AAA')])
# run
msa_1.add_gap_to_sequence_at_index(seq_index=0, gap_position=0)
msa_2.add_gap_to_sequence_at_index(seq_index=0, gap_position=2)
msa_3.add_gap_to_sequence_at_index(seq_index=0, gap_position=3)
msa_4.add_gap_to_sequence_at_index(seq_index=0, gap_position=1)
# check
self.assertEqual([('seq1', '-A-')],
msa_1.decode_alignment_as_list_of_pairs())
self.assertEqual([('seq1', '-A-')],
msa_2.decode_alignment_as_list_of_pairs())
self.assertEqual([('seq1', 'A-C-')],
msa_3.decode_alignment_as_list_of_pairs())
self.assertEqual([('seq1', 'A-AA')],
msa_4.decode_alignment_as_list_of_pairs())
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
