def test_filter_gap_high_entropy_low(self):
result = filter_positions(self.alignment_with_gaps,
self.maximum_gap_frequency_100,
self.maximum_position_entropy_10)
aln = Alignment([BiologicalSequence('A-', id="seq1"),
BiologicalSequence('A-', id="seq2"),
BiologicalSequence('A-', id="seq3"),
BiologicalSequence('A-', id="seq4")])
self.assertEqual(result, aln)
def test_filter_gap_high_entropy_low(self):
result = filter_positions(self.alignment_with_gaps,
self.maximum_gap_frequency_100,
self.maximum_position_entropy_10)
aln = Alignment([
BiologicalSequence('A-', id="seq1"),
BiologicalSequence('A-', id="seq2"),
BiologicalSequence('A-', id="seq3"),
BiologicalSequence('A-', id="seq4")
])
self.assertEqual(result, aln)
def test_filter_gap_high_entropy_low(self):
result = filter_positions(self.alignment_with_gaps,
self.maximum_gap_frequency_100,
self.maximum_position_entropy_10)
aln = TabularMSA([
DNA('A-', metadata={'id': "seq1"}),
DNA('A-', metadata={'id': "seq2"}),
DNA('A-', metadata={'id': "seq3"}),
DNA('A-', metadata={'id': "seq4"})
])
self.assertEqual(str(result), str(aln))
def filter_alignment_positions(aligned_sequences_file: AlignedDNAFASTAFormat,
maximum_gap_frequency: str,
maximum_position_entropy: str) -> \
AlignedDNAFASTAFormat:
aligned_sequences_fh = aligned_sequences_file.open()
fasta_file = AlignedDNAFASTAFormat()
skbio.write(filter_positions(aligned_sequences_fh, maximum_gap_frequency,
maximum_position_entropy),
into=str(fasta_file),
format='fasta')
return fasta_file
def test_filter_gap_high_entropy_low(self):
result = filter_positions(
self.alignment_with_gaps, self.maximum_gap_frequency_100, self.maximum_position_entropy_10
)
aln = TabularMSA(
[
DNA("A-", metadata={"id": "seq1"}),
DNA("A-", metadata={"id": "seq2"}),
DNA("A-", metadata={"id": "seq3"}),
DNA("A-", metadata={"id": "seq4"}),
]
)
self.assertEqual(str(result), str(aln))
