def test_search_multi():
rsidlist = ['rs60995877']
vcffile = data_file('chr9-multi.vcf.gz')
idxfile = data_file('chr9-multi.rsidx')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile))
assert len(outlines) == 7
for line in outlines:
assert line.split('\t')[2] == 'rs60995877'
def test_search_overlapping_variants(doheader, numlines):
rsidlist = ['rs8051733']
vcffile = data_file('overlap.vcf.gz')
idxfile = data_file('overlap.sqlite3')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile, doheader))
assert len(outlines) == numlines
assert '\trs8051733\t' in outlines[-1]
assert '\trs967556605\t' not in outlines[-1]
def test_search_multiple_rsids_single_query():
for rsidlist in [['rs72634902'], ['rs145742571']]:
vcffile = data_file('multiple_id.vcf.gz')
idxfile = data_file('multiple_id.rsidx')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile))
assert len(outlines) == 1
assert outlines[0].startswith(
'1\t1900106\trs72634902;rs145742571\tT\tC,TCTC')
conn.close()
def test_search_missing_rsid(capsys):
rsidlist = [123456789]
vcffile = data_file('chr17-sample.vcf.gz')
idxfile = data_file('chr17-sample.rsidx')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile))
assert len(outlines) == 0
conn.close()
terminal = capsys.readouterr()
assert '[rsidx::search] WARNING: no rsID matches' in terminal.err
def test_search_stdout(capsys):
arglist = [
'search',
data_file('chr17-sample.vcf.gz'),
data_file('chr17-sample.rsidx'), 'rs1472751972', 'rs1287502205',
'rs897983471', 'rs1172219431', 'rs189123651'
]
args = rsidx.cli.get_parser().parse_args(arglist)
rsidx.search.main(args)
terminal = capsys.readouterr()
outlines = terminal.out.strip().split('\n')
assert len(outlines) == 5
def test_search_bad_rsids():
rsidlist = [
'rs538736078', # replaced by . in VCF
'rs547329663', # replaced by bogus ID in VCF
'rs1440788236', # valid RSID not present in VCF
'rs1234497371', # valid RSID present in VCF
]
vcffile = data_file('chr4-sample-corrupted-ids.vcf.gz')
idxfile = data_file('chr4-sample-corrupted-ids.rsidx')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile))
assert len(outlines) == 1
assert outlines[0].startswith('4\t218446\trs1234497371\tC\tCA,CAA')
conn.close()
def test_search(rsidlist):
vcffile = data_file('chr17-sample.vcf.gz')
idxfile = data_file('chr17-sample.rsidx')
conn = sqlite3.connect(idxfile)
outlines = list(rsidx.search.search(rsidlist, conn, vcffile))
assert len(outlines) == 5
outdata = [line.split('\t')[:5] for line in outlines]
assert sorted(outdata) == sorted(
[['17', '944196', 'rs182553373', 'G', 'A'],
['17', '611663', 'rs544992196', 'T', 'C'],
['17', '1946968', 'rs1245348147', 'T', 'C'],
['17', '567599', 'rs1335948438', 'C', 'T'],
['17', '374561', 'rs1440788236', 'G', 'T']])
conn.close()
def test_search_cli(doheader, numlines, suffix):
with NamedTemporaryFile(suffix=suffix) as outfile:
arglist = [
'search',
data_file('chr17-sample.vcf.gz'),
data_file('chr17-sample.rsidx'), '--out', outfile.name,
'rs1472751972', 'rs1287502205', 'rs897983471', 'rs1172219431',
'rs189123651'
]
args = rsidx.cli.get_parser().parse_args(arglist)
args.header = doheader
rsidx.search.main(args)
with rsidx.open(outfile.name, 'r') as fh:
outlines = fh.read().strip().split('\n')
assert len(outlines) == numlines
def test_index_force_reindex(capsys):
with TempFileName(suffix='.rsidx') as idxfile:
arglist = ['index', '--force', data_file('chr9-multi.vcf.gz'), idxfile]
args = rsidx.cli.get_parser().parse_args(arglist)
rsidx.index.main(args)
rsidx.index.main(args)
terminal = capsys.readouterr()
assert ', overwriting' in terminal.err
def test_index_no_force_reindex(capsys):
with TempFileName(suffix='.rsidx') as idxfile:
arglist = ['index', data_file('chr9-multi.vcf.gz'), idxfile]
args = rsidx.cli.get_parser().parse_args(arglist)
rsidx.index.main(args)
with pytest.raises(SystemExit):
rsidx.index.main(args)
terminal = capsys.readouterr()
assert ', stubbornly refusing to proceed' in terminal.err
def test_index_multi(capsys):
vcffile = data_file('chr9-multi.vcf.gz')
with TempFileName(suffix='.rsidx') as idxfile, rsidx.open(vcffile,
'r') as vcffh:
with sqlite3.connect(idxfile) as dbconn:
rsidx.index.index(dbconn, vcffh)
arglist = ['search', vcffile, idxfile, 'rs60995877']
args = rsidx.cli.get_parser().parse_args(arglist)
rsidx.search.main(args)
terminal = capsys.readouterr()
assert terminal.out.count('\trs60995877\t') == 7
def test_index_bogus_rsids():
with NamedTemporaryFile(suffix='.sqlite3') as db:
with sqlite3.connect(db.name) as dbconn:
vcffile = data_file('chr4-sample-corrupted-ids.vcf.gz')
with rsidx.open(vcffile, 'r') as vcffh:
rsidx.index.index(dbconn, vcffh)
c = dbconn.cursor()
query = ('SELECT * FROM rsid_to_coord WHERE rsid IN '
'(538736078, 547329663, 1440788236, 1234497371)')
results = list(c.execute(query))
assert results == [(1234497371, '4', 218446)]
def test_index_cli(mainfunc):
with TempFileName(suffix='.rsidx') as idxfile:
arglist = ['index', data_file('chr17-sample.vcf.gz'), idxfile]
args = rsidx.cli.get_parser().parse_args(arglist)
mainfunc(args)
conn = sqlite3.connect(idxfile)
c = conn.cursor()
query = ('SELECT * FROM rsid_to_coord WHERE rsid IN '
'(548749810, 956322221)')
results = list(c.execute(query))
assert sorted(results) == sorted([(548749810, '17', 1098730),
(956322221, '17', 1227227)])
def test_index_multi_rsids():
with NamedTemporaryFile(suffix='.sqlite3') as db:
with sqlite3.connect(db.name) as dbconn:
vcffile = data_file('multiple_id.vcf.gz')
with rsidx.open(vcffile, 'r') as vcffh:
rsidx.index.index(dbconn, vcffh)
c = dbconn.cursor()
query = ('SELECT * FROM rsid_to_coord WHERE rsid IN '
'(72634902, 145742571)')
results = list(c.execute(query))
assert sorted(results) == sorted([(72634902, '1', 1900106),
(145742571, '1', 1900106)])
def test_index(cachesize, mmapsize):
with NamedTemporaryFile(suffix='.sqlite3') as db:
with sqlite3.connect(db.name) as dbconn:
with rsidx.open(data_file('chr17-sample.vcf.gz'), 'r') as vcffh:
rsidx.index.index(dbconn,
vcffh,
cache_size=cachesize,
mmap_size=mmapsize,
logint=10)
c = dbconn.cursor()
query = ('SELECT * FROM rsid_to_coord WHERE rsid IN '
'(1238461543, 1472751972)')
results = list(c.execute(query))
assert sorted(results) == sorted([(1238461543, '17', 624973),
(1472751972, '17', 132359)])
