def test_build38_exact_search_variants(self, v600e):
query = CoordinateQuery('7', 140753336, 140753336, 'T', 'A', 'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert len(search_results) == 1
assert search_results[0] == v600e
query = CoordinateQuery('7', 140753336, 140753337, 'TT', 'AC',
'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert len(search_results) == 1
assert search_results[0].id == 563
query = CoordinateQuery('3', 10146548, 10146549, 'C', None, 'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert len(search_results) == 1
assert search_results[0].id == 1918
query = CoordinateQuery('3', 10146618, 10146618, None, 'G', 'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert len(search_results) == 1
assert search_results[0].id == 2042
def annotate_vcf(input_vcf, output_vcf, reference, include_status):
"""Annotate a VCF with information from CIViC"""
reader = vcfpy.Reader.from_path(input_vcf)
new_header = reader.header.copy()
new_header.add_info_line(
OrderedDict([('ID', 'CIVIC'), ('Number', '.'), ('Type', 'String'),
('Description', VCFWriter.CSQ_DESCRIPTION)]))
writer = vcfpy.Writer.from_path(output_vcf, new_header)
for entry in reader:
for alt in entry.ALT:
position = entry.POS
ref = entry.REF
alt = alt.value
if len(ref) == 1 and len(alt) == 1:
start = position
end = position
else:
if len(ref) == len(alt):
start = position
end = position + len(ref) - 1
else:
alt = alt[1:]
ref = ref[1:]
if len(ref) > len(alt):
start = position + 1
end = start + len(ref) - 1
if alt == '':
alt = None
else:
start = position
if ref == '':
ref = None
end = start + 1
else:
end = start + len(ref) - 1
query = CoordinateQuery(entry.CHROM, start, end, alt, ref,
reference)
variants = civic.search_variants_by_coordinates(
query, search_mode='exact')
if variants is not None:
if len(variants) == 1:
csq = variants[0].csq(include_status)
if len(csq) > 0:
entry.INFO['CIVIC'] = variants[0].csq(include_status)
elif len(variants) > 1:
print(
"More than one variant found for start {} stop {} ref {} alt {}. CIViC Variants IDs: {}"
.format(
start, end, ref, alt,
",".join(list(map(lambda v: str(v.id),
variants)))))
writer.write_record(entry)
writer.close()
reader.close()
def test_errors(self):
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, 'T', 'A')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='wrong_mode')
assert "unexpected search mode" in str(context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140753336, 140753336, '*', 'A',
'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "Can't use wildcard when searching for non-GRCh37 coordinates" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140753336, 140753336, None, None,
'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "alt or ref required for non-GRCh37 coordinate queries" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140753336, 140753336, 'T', 'A',
'GRCh38')
search_results = civic.search_variants_by_coordinates(
query, search_mode='any')
assert "Only exact search mode is supported for non-GRCh37 coordinate queries" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, '-', 'A')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "Unexpected alt `-` in coordinate query. Did you mean `None`?" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, 'T', '-')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "Unexpected ref `-` in coordinate query. Did you mean `None`?" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, '-', 'A',
'GRCh38')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "Unexpected alt `-` in coordinate query. Did you mean `None`?" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, 'T', '-',
'GRCh38')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert "Unexpected ref `-` in coordinate query. Did you mean `None`?" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, '-', 'A')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert "Unexpected alt `-` in coordinate query. Did you mean `None`?" in str(
context.value)
with pytest.raises(ValueError) as context:
query = CoordinateQuery('7', 140453136, 140453136, 'T', '-')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert "Unexpected ref `-` in coordinate query. Did you mean `None`?" in str(
context.value)
def test_build36_exact_search_variants(self, v600e):
query = CoordinateQuery('7', 140099605, 140099605, 'T', 'A', 'NCBI36')
search_results = civic.search_variants_by_coordinates(
query, search_mode='exact')
assert len(search_results) == 1
assert search_results[0] == v600e
def test_single_and_bulk_exact_return_same_variants(self):
query = CoordinateQuery('7', 140453136, 140453136, 'T', '*')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('7', 140453136, 140453136, 'T', 'A')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('7', 140453136, 140453136, 'T', None)
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 0
assert len(variants_bulk) == 0
query = CoordinateQuery('7', 140453136, 140453137, 'TT', '*')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('7', 140453136, 140453137, 'TT', 'AC')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('7', 140453136, 140453137, 'TT', None)
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 0
assert len(variants_bulk) == 0
query = CoordinateQuery('3', 10183706, 10183706, None, 'C')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('3', 10183706, 10183706, 'T', 'C')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
assert len(variants_single) == 1
assert len(variants_bulk[query]) == 1
assert hash(variants_single[0]) == variants_bulk[query][0].v_hash
query = CoordinateQuery('3', 10183706, 10183706, '*', 'C')
variants_single = civic.search_variants_by_coordinates(
query, search_mode='exact')
variants_bulk = civic.bulk_search_variants_by_coordinates(
[query], search_mode='exact')
variants_single = list(map(lambda v: hash(v), variants_single))
variants_bulk = list(map(lambda v: v.v_hash, variants_bulk[query]))
assert len(variants_single) == 2
assert len(variants_bulk) == 2
assert sorted(variants_single) == sorted(variants_bulk)
assert sorted(variants_single) == sorted(variants_bulk)