def elongate_data():
seq_id = 'test1'
test_ann = [
gff.Annotation(seq_id=seq_id, start=1, end=10),
gff.Annotation(seq_id=seq_id, start=10, end=15),
gff.Annotation(seq_id=seq_id, start=16, end=18),
]
return test_ann
def test_Annotation_get_aa_seq():
ann = gff.Annotation(start=1, end=40, strand='+', phase=0)
seq = 'ACTG' * 10
assert ann.get_aa_seq(seq) == sequence.translate_sequence(
ann.get_nuc_seq(seq, reverse=True),
start=0,
reverse=False
)
def test_Annotation_get_aa_seq__snp():
ann = gff.Annotation(start=1, end=40, strand='+', phase=1)
seq = 'ACTG' * 10
assert ann.get_aa_seq(seq, start=None, snp=(1, 'C')) == sequence.translate_sequence(
ann.get_nuc_seq(seq, reverse=True, snp=(1, 'C')),
start=ann.phase,
reverse=False
)
def test_genomicrange_contains_annotation1(coords1, coords2, result):
gen_range1 = gff.GenomicRange(start=coords1[0], end=coords1[1])
assert (gff.Annotation(start=coords2[0], end=coords2[1]) in gen_range1) == result
def test_Annotation_is_syn__start3_2():
# first position on reference, third codon, second base in codon, phase=1
ann = gff.Annotation(start=1, end=40, strand='+', phase=1)
seq = 'ACTG' * 10
assert not ann.is_syn(seq, 9, 'C', start=None)
def test_Annotation_is_syn__start2_1():
# first position on reference, second codon, first base in codon, phase=1
ann = gff.Annotation(start=1, end=40, strand='+', phase=1)
seq = 'ACTG' * 10
assert not ann.is_syn(seq, 6, 'A', start=None)
def test_Annotation_is_syn4_3():
# first position on reference, third codon, third base in codon
ann = gff.Annotation(start=1, end=40, strand='+', phase=0)
seq = 'ACTG' * 10
assert not ann.is_syn(seq, 9, 'T')
def test_Annotation_is_syn2_3():
# second position on reference, third base in codon
ann = gff.Annotation(start=2, end=40, strand='+', phase=0)
seq = 'ACTG' * 10
assert ann.is_syn(seq, 4, 'A')
def test_Annotation_is_syn1_1():
ann = gff.Annotation(start=1, end=40, strand='+', phase=0)
seq = 'ACTG' * 10
assert not ann.is_syn(seq, 1, 'C')
def test_Annotation_taxon_id_set1():
ann = gff.Annotation()
ann.taxon_id = '10'
assert ann.taxon_id == 10
def test_annotation_length1():
ann = gff.Annotation(start=1, end=10)
assert len(ann) == 10
def test_Annotation_get_nuc_seq_snp():
ann = gff.Annotation(start=2, end=39, strand='+')
seq = 'ACTG' * 10
ann.get_nuc_seq(seq, snp=(2, 'C')) == sequence.get_variant_sequence(seq[1:-1], (2, 'C'))
def test_Annotation_get_nuc_seq_reverse2():
ann = gff.Annotation(start=2, end=39, strand='-')
seq = 'ACTG' * 10
assert ann.get_nuc_seq(seq, reverse=True) == sequence.reverse_complement(seq[1:-1])
def test_Annotation_get_nuc_seq_reverse1():
ann = gff.Annotation(start=1, end=40, strand='-')
seq = 'ACTG' * 10
assert ann.get_nuc_seq(seq) == seq
def test_Annotation_get_nuc_seq3():
ann = gff.Annotation(start=2, end=39, strand='+')
seq = 'ACTG' * 10
assert ann.get_nuc_seq(seq) == seq[1:-1]
def test_Annotation_taxon_id_set3():
ann = gff.Annotation()
ann.taxon_id = 10.0
assert ann.taxon_id == 10
def test_genomicrange_contains_annotation(start1, end1, start2, end2, result):
gen_range1 = gff.GenomicRange(start=start1, end=end1)
assert (gff.Annotation(start=start2, end=end2) in gen_range1) == result
def test_Annotation_set_mapping2():
ann = gff.Annotation()
ann.set_mapping('ko', ['test1', 'test2'])
assert ann.attr['map_KO'] == 'test1,test2'
def test_Annotation_taxon_id_get1():
ann = gff.Annotation()
assert ann.taxon_id is None
def test_annotation_length2():
ann = gff.Annotation(start=1, end=10)
assert ann.length == 10
def test_Annotation_get_mappings1():
ann = gff.Annotation()
ann.set_mapping('ko', ['test'])
mappings = ann.get_mappings()
assert mappings['ko'] == ['test']
