def test_build_popfile(tmp_path):
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
fp.build_files()
assert fp.files['popfile'] is None
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['popfile'] == os.path.join(
str(tmp_path), fp.get_filename('.popfile'))
opts = admixture_option_parser().parse_args(
['--popfile', 'test'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['popfile'] == \
os.path.join(os.getcwd(), 'test.popfile')
opts = admixture_option_parser().parse_args(
['--popfile', 'test', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['popfile'] == \
os.path.join(str(tmp_path), 'test.popfile')
def test_build_f4dstat(tmp_path):
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
fp.build_files()
files = ['f4dstat', 'eigen', 'snp', 'ind']
for f in files:
assert fp.files[f] is None
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
outbases = ['parfile.F4stat', 'eigenstratgeno', 'snp', 'ind']
for f, ob in zip(files, outbases):
assert fp.files[f] == os.path.join(
str(tmp_path), fp.get_f4dstat_filename(ob))
opts = admixture_option_parser().parse_args(
['--f4dstat', 'test'])
fp = file_printer(opts)
fp.build_files()
outbases = ['parfile.F4stat.{}.gz', 'eigenstratgeno.{}.gz',
'snp.{}.gz', 'ind.{}.gz']
for f, ob in zip(files, outbases):
assert fp.files[f] == \
os.path.join(os.getcwd(), ob.format('test'))
opts = admixture_option_parser().parse_args(
['--f4dstat', 'test', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
for f, ob in zip(files, outbases):
assert fp.files[f] == \
os.path.join(str(tmp_path), ob.format('test'))
def test_print_debug():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
model = Demography_Models.Base_demography(opts)
with file_printer(opts) as fp:
fp.writers['debug'] = output
fp.print_debug(model)
def main():
options = admixture_option_parser().parse_args()
model = get_model(options)
with file_printer(options) as printer:
printer.print_options()
printer.print_debug(model)
if printer.single_simulation_needed():
simulation = model.simulate(replicates=1)
tree_sequence = next(simulation)
printer.print_popfile(model, tree_sequence)
printer.print_vcf(tree_sequence)
printer.print_pi(tree_sequence, model.get_sample_indices(),
model.get_population_map())
if printer.haplo_needed():
haplotype_entry_list = get_haplo_entries(
tree_sequence, options)
printer.print_haplo(haplotype_entry_list)
if printer.ils_needed():
haplotype_entry_list = get_haplo_entries(tree_sequence,
options,
isILS=True)
printer.print_ils(haplotype_entry_list)
# simulate with 20 replicate for F4Dstat
if printer.f4dstat_needed():
simulation = model.simulate(replicates=20)
write_f4dstats(simulation, printer, model)
def test_write_to():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
fp.writers['test'] = output
fp.write_to('test', 'testing')
assert output.getvalue() == 'testing'
def test_exeception_options():
# too many flags set
with pytest.raises(ValueError) as e:
opts = admixture_option_parser().parse_args(
['--options', '--haplo'])
file_printer(opts)
assert 'Expected at most one output to stdout, got 2 instead.' in str(e)
with pytest.raises(ValueError) as e:
opts = admixture_option_parser().parse_args(
['--vcf', 'vc', '--haplo', '--options', '--debug'])
file_printer(opts)
assert 'Expected at most one output to stdout, got 3 instead.' in str(e)
with pytest.raises(ValueError) as e:
opts = admixture_option_parser().parse_args(
['--haplo', '--options', '--debug', '--pi'])
file_printer(opts)
assert 'Expected at most one output to stdout, got 4 instead.' in str(e)
def test_print_options():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
fp.writers['options'] = output
fp.print_options()
assert 'pop: nonAfr' in output.getvalue()
assert 's_n1: 2' in output.getvalue()
output = io.StringIO()
opts = admixture_option_parser().parse_args(
'-p AFR --Neand1_sample_size 3'.split())
with file_printer(opts) as fp:
fp.writers['options'] = output
fp.print_options()
assert 'pop: AFR' in output.getvalue()
assert 's_n1: 3' in output.getvalue()
def test_build_pi(tmp_path):
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] is None
opts = admixture_option_parser().parse_args(['--pi'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--pi', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] == os.path.join(
str(tmp_path), 'pi.txt')
opts = admixture_option_parser().parse_args(
['--pi', 'test.txt'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] == os.path.join(os.getcwd(), 'test.txt')
opts = admixture_option_parser().parse_args(
['--pi', 'test.txt', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['pi'] == os.path.join(str(tmp_path), 'test.txt')
def test_build_out_dir(tmp_path):
# default set to wd
opts = admixture_option_parser().parse_args(
['--haplo', 'hap'])
fp = file_printer(opts)
fp.build_out_dir()
assert fp.out_dir == os.getcwd()
# new dir (pytest starts with existing dir)
os.rmdir(tmp_path)
assert not os.path.exists(tmp_path)
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path), '--haplo', 'hap'])
fp = file_printer(opts)
fp.build_out_dir()
assert os.path.exists(tmp_path)
# redoing with existing dir should not fail
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path), '--haplo', 'hap'])
fp = file_printer(opts)
fp.build_out_dir()
def test_print_vcf():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
# when writer is still none
fp.print_vcf(None)
model = Demography_Models.Base_demography(opts)
ts = next(model.simulate(1))
fp.writers['vcf'] = output
fp.print_vcf(ts)
output = output.getvalue().split('\n')
assert "##source=msprime 0.6.1" == output[1]
assert len(output[5].split('\t')) == 1021
def test_print_popfile():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
# when popfile writer is still none
fp.print_popfile(None, None)
model = Demography_Models.Base_demography(opts)
ts = next(model.simulate(1))
fp.writers['popfile'] = output
fp.print_popfile(model, ts)
output = output.getvalue().split('\n')
assert "samp\tpop\tsuper_pop" == output[0]
assert "msp_0\tNeand1\tNeand1" == output[1]
assert "msp_1011\tDeni\tDeni" == output[-2]
def test_print_haplo():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
# when writer is still none
fp.print_haplo(None)
assert fp.haplo_needed() is False
haplos = {2: [[100, 300], [200, 400]]}
fp.writers['haplo'] = output
assert fp.haplo_needed() is True
fp.print_haplo(haplos)
output = output.getvalue().split('\n')
assert output[0] == '1\t100\t200\t2'
assert output[1] == '1\t300\t400\t2'
def test_build_debug(tmp_path):
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == sys.stdout
opts = admixture_option_parser().parse_args(['--debug'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--debug', '--vcf', 'vcfbase'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--debug', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == os.path.join(str(tmp_path), 'debug.txt')
opts = admixture_option_parser().parse_args(
['--haplo'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] is None
opts = admixture_option_parser().parse_args(
['--debug', 'test.txt'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['debug'] == os.path.join(os.getcwd(), 'test.txt')
opts = admixture_option_parser().parse_args(
['--debug', 'test.txt', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
def test_print_f4dstat():
output = io.StringIO()
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
fp.print_f4dstat()
fp.writers['f4dstat'] = output
fp.files['eigen'] = 'eigen'
fp.files['snp'] = 'snp'
fp.files['ind'] = 'ind'
fp.print_f4dstat()
output = output.getvalue().split('\n')
assert 'genotypename: eigen' == output[0]
assert 'snpname: snp' == output[1]
assert 'indivname: ind' == output[2]
assert 'popfilename: sim.popfile_F4stat' == output[3]
def test_options():
# single flags set
opts = admixture_option_parser().parse_args(['--haplo'])
fp = file_printer(opts)
assert fp.files['haplo'] == '*'
assert fp.files['debug'] is None
opts = admixture_option_parser().parse_args(['--options'])
fp = file_printer(opts)
assert fp.files['options'] == '*'
assert fp.files['debug'] is None
# flag set with filename on others
opts = admixture_option_parser().parse_args(
['--debug', '--vcf', 'vcf.out'])
fp = file_printer(opts)
assert fp.files['options'] is None
assert fp.files['debug'] == '*'
assert fp.files['vcf'] == 'vcf.out'
opts = admixture_option_parser().parse_args(
['--options', '--f4dstat', 'f4'])
fp = file_printer(opts)
assert fp.files['haplo'] is None
assert fp.files['f4dstat'] == 'f4'
assert fp.files['options'] == '*'
# outdir set ok
opts = admixture_option_parser().parse_args(
['--out-dir', 'outd'])
fp = file_printer(opts)
assert fp.files['haplo'] is None
assert fp.out_dir == 'outd'
# note this is valid but silly, no output will go to outdir
opts = admixture_option_parser().parse_args(
['--out-dir', 'outd', '--haplo'])
fp = file_printer(opts)
assert fp.out_dir == 'outd'
assert fp.files['haplo'] == '*'
opts = admixture_option_parser().parse_args(
['--out-dir', 'outd', '--haplo', 'hap'])
fp = file_printer(opts)
assert fp.out_dir == 'outd'
assert fp.files['haplo'] == 'hap'
def test_print_pi():
output = io.StringIO()
opts = admixture_option_parser().parse_args('-r 216'.split())
with file_printer(opts) as fp:
# when writer is still none
fp.print_pi(None, None, None)
model = Demography_Models.Base_demography(opts)
ts = next(model.simulate(1))
fp.writers['pi'] = output
fp.print_pi(ts, model.get_sample_indices(),
model.get_population_map())
print(output.getvalue())
lines = output.getvalue().split('\n')
assert lines[0].split() == 'AF EU AS AF-EU AF-AS EU-AS'.split()
tokens = lines[1].split()
assert tokens[0] == '0.27959' # vcftools 0.27937
assert tokens[1] == '0.32405' # vcftools 0.323697
assert tokens[2] == '0.34194' # vcftools 0.341196
assert tokens[3] == '0.2119' # vcftools 0.211696
assert tokens[4] == '0.45429' # vcftools 0.453715
assert tokens[5] == '0.19144' # vcftools 0.191207
def test_open_writers(tmp_path):
opts = admixture_option_parser().parse_args([])
with file_printer(opts) as fp:
for k, v in fp.writers.items():
if v is not None:
assert k == 'debug'
assert fp.writers['debug'] == sys.stdout
assert fp.haplo_needed() is False
assert fp.f4dstat_needed() is False
singleopts = ['debug', 'options', 'haplo']
for o in singleopts:
opts = admixture_option_parser().parse_args(['--' + o])
with file_printer(opts) as fp:
for k, v in fp.writers.items():
if v is not None:
assert k == o
assert fp.writers[o] == sys.stdout
assert fp.single_simulation_needed() is (o == 'haplo')
assert fp.haplo_needed() is (o == 'haplo')
assert fp.f4dstat_needed() is False
opts = admixture_option_parser().parse_args(['--haplo', 'test.gz',
'--out-dir', str(tmp_path)])
opts = admixture_option_parser().parse_args(['--haplo', 'test.vcf',
'--out-dir', str(tmp_path)])
opts = admixture_option_parser().parse_args(['--vcf', 'test',
'--out-dir', str(tmp_path)])
with file_printer(opts) as fp:
for k, v in fp.writers.items():
if v is not None:
assert k == 'vcf' or k == 'popfile'
opts = admixture_option_parser().parse_args(['--f4dstat', 'test',
'--out-dir', str(tmp_path)])
with file_printer(opts) as fp:
for k, v in fp.writers.items():
if v is not None:
assert k == 'f4dstat' \
or k == 'eigen' \
or k == 'snp' \
or k == 'ind'
assert fp.haplo_needed() is False
assert fp.f4dstat_needed() is True
def test_build_haplo(tmp_path):
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] is None
opts = admixture_option_parser().parse_args(['--haplo'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--haplo', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] == sys.stdout
opts = admixture_option_parser().parse_args(
['--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] == os.path.join(
str(tmp_path),
fp.get_filename('.bed.merged.gz'))
opts = admixture_option_parser().parse_args(
['--haplo', 'test.txt'])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] == os.path.join(os.getcwd(), 'test.txt')
opts = admixture_option_parser().parse_args(
['--haplo', 'test.txt', '--out-dir', str(tmp_path)])
fp = file_printer(opts)
fp.build_files()
assert fp.files['haplo'] == os.path.join(str(tmp_path), 'test.txt')
def test_defaults():
opts = admixture_option_parser().parse_args([])
fp = file_printer(opts)
for f in fp.files.values():
assert f is None
assert fp.out_dir is None
