def test_missing_format_keys(self):
values = [{}, {"format_name": ""}, {"format_version": ""}]
for attrs in values:
self.write_file(attrs, self.path)
with self.assertRaises(exceptions.FileFormatError):
tszip.decompress(self.path)
with self.assertRaises(exceptions.FileFormatError):
tszip.print_summary(self.path)
def test_bad_format_name(self):
for bad_name in ["", "xyz", [1234]]:
self.write_file(
{"format_name": bad_name, "format_version": [1, 0]}, self.path
)
with self.assertRaises(exceptions.FileFormatError):
tszip.decompress(self.path)
with self.assertRaises(exceptions.FileFormatError):
tszip.print_summary(self.path)
def test_format_too_new(self):
self.write_file({
"format_name": "tszip",
"format_version": [2, 0]
}, self.path)
with self.assertRaises(exceptions.FileFormatError):
tszip.decompress(self.path)
with self.assertRaises(exceptions.FileFormatError):
tszip.print_summary(self.path)
def test_keep(self):
self.assertTrue(self.trees_path.exists())
self.run_tszip([str(self.trees_path), "--keep"])
self.assertTrue(self.trees_path.exists())
outpath = pathlib.Path(str(self.trees_path) + ".tsz")
self.assertTrue(outpath.exists())
ts = tszip.decompress(outpath)
self.assertEqual(ts.tables, self.ts.tables)
def test_suffix(self):
self.assertTrue(self.trees_path.exists())
self.run_tszip([str(self.trees_path), "-S", ".XYZasdf"])
self.assertFalse(self.trees_path.exists())
outpath = pathlib.Path(str(self.trees_path) + ".XYZasdf")
self.assertTrue(outpath.exists())
ts = tszip.decompress(outpath)
self.assertEqual(ts.tables, self.ts.tables)
def run_decompress(args):
logger.info("Decompressing {}".format(args.file))
if not args.file.endswith(".zarr"):
raise ValueError("Compressed file must have .zarr suffix")
ts = tszip.decompress(args.file)
outfile = args.file[:-5]
logger.info("Writing to {}".format(outfile))
ts.dump(outfile)
def verify(self, ts):
with tempfile.TemporaryDirectory() as tmpdir:
path = pathlib.Path(tmpdir) / "treeseq.tsz"
tszip.compress(ts, path)
other_ts = tszip.decompress(path)
G1 = ts.genotype_matrix()
G2 = other_ts.genotype_matrix()
self.assertTrue(np.array_equal(G1, G2))
def test_overwrite(self):
self.assertTrue(self.trees_path.exists())
outpath = pathlib.Path(str(self.trees_path) + ".tsz")
outpath.touch()
self.assertTrue(self.trees_path.exists())
self.run_tszip([str(self.trees_path), "--force"])
self.assertFalse(self.trees_path.exists())
self.assertTrue(outpath.exists())
ts = tszip.decompress(outpath)
self.assertEqual(ts.tables, self.ts.tables)
def test_variants_only(self):
self.assertTrue(self.trees_path.exists())
self.run_tszip([str(self.trees_path), "--variants-only"])
self.assertFalse(self.trees_path.exists())
outpath = pathlib.Path(str(self.trees_path) + ".tsz")
self.assertTrue(outpath.exists())
ts = tszip.decompress(outpath)
self.assertNotEqual(ts.tables, self.ts.tables)
G1 = ts.genotype_matrix()
G2 = self.ts.genotype_matrix()
self.assertTrue(np.array_equal(G1, G2))
def run_decompress(args):
setup_logging(args)
for file_arg in args.files:
logger.info("Decompressing {}".format(file_arg))
if not file_arg.endswith(args.suffix):
exit("Compressed file must have '{}' suffix".format(args.suffix))
infile = pathlib.Path(file_arg)
outfile = pathlib.Path(file_arg[:-len(args.suffix)])
check_output(outfile, args)
with check_load_errors(file_arg):
ts = tszip.decompress(file_arg)
ts.dump(str(outfile))
logger.info("Wrote {}".format(outfile))
remove_input(infile, args)
def run_decompress(args):
setup_logging(args)
for file_arg in args.files:
logger.info(f"Decompressing {file_arg}")
if not file_arg.endswith(args.suffix):
exit(f"Compressed file must have '{args.suffix}' suffix")
infile = pathlib.Path(file_arg)
if args.stdout:
args.keep = True
outfile = sys.stdout
else:
outfile = pathlib.Path(file_arg[:-len(args.suffix)])
check_output(outfile, args)
with check_load_errors(file_arg):
ts = tszip.decompress(file_arg)
ts.dump(outfile)
logger.info(f"Wrote {outfile}")
remove_input(infile, args)
def verify(self, ts):
if ts.num_migrations > 0:
raise unittest.SkipTest("Migrations not supported")
with tempfile.TemporaryDirectory() as tmpdir:
path = pathlib.Path(tmpdir) / "treeseq.tsz"
tszip.compress(ts, path, variants_only=True)
other_ts = tszip.decompress(path)
self.assertEqual(ts.num_sites, other_ts.num_sites)
for var1, var2 in zip(ts.variants(), other_ts.variants()):
self.assertTrue(np.array_equal(var1.genotypes, var2.genotypes))
self.assertEqual(var1.site.position, var2.site.position)
self.assertEqual(var1.alleles, var2.alleles)
# Populations, individuals and sites should be untouched if there are no
# unreachable individuals.
t1 = ts.tables
t2 = other_ts.tables
self.assertEqual(t1.sequence_length, t2.sequence_length)
self.assertEqual(t1.populations, t2.populations)
self.assertEqual(t1.individuals, t2.individuals)
self.assertEqual(t1.sites, t2.sites)
# We should be adding an extra provenance record in here due to simplify.
self.assertEqual(len(t1.provenances), len(t2.provenances) - 1)
# In[12]:
arg_list = sys.argv
model = arg_list[1] + '_'
list_ts_files = [
x for x in os.listdir('tree_seq_files/') if x.startswith(model)
]
n = len(list_ts_files)
# ## 1) Compute statistics from list of .ts files
# In[13]:
# iterate through that list, getting stats for every file object
for file in list_ts_files:
ts = tszip.decompress('tree_seq_files/' + file)
sample_list = make_sample_list(ts)
tajimas_pi.append(ts.diversity(sample_sets=sample_list))
D_DenAnc.append(D_AncEst(ts, "DEN"))
D_NeaAnc.append(D_AncEst(ts, "NEA"))
f4_DenAnc_a.append(f4_DenAncEst(ts, "a"))
f4_DenAnc_b.append(f4_DenAncEst(ts, "b"))
f4_Est_c.append(f4_AncEst(ts, "c"))
f4_Est_d.append(f4_AncEst(ts, "d"))
f4_Est_e.append(f4_AncEst(ts, "e"))
mig_segs_DEN_AYT, mig_segs_DEN_PAP, mig_segs_NEA_AYT, mig_segs_NEA_PAP = get_mig_segs(
ts)
mean_tractL_DEN.append(
np.array([np.mean(mig_segs_DEN_PAP),
np.mean(mig_segs_DEN_AYT)]))
mean_tractL_NEA.append(
def test_load_dir(self):
with self.assertRaises(OSError):
tszip.decompress(self.path.parent)
def test_missing_file(self):
path = "/no/such/file"
with self.assertRaises(FileNotFoundError):
tszip.decompress(path)
def test_wrong_format(self):
for contents in ["", "1234", "X" * 1024]:
with open(str(self.path), "w") as f:
f.write(contents)
with self.assertRaises(exceptions.FileFormatError):
tszip.decompress(self.path)
def test_decompress(self):
files = pathlib.Path(__file__).parent / "files"
ts = tszip.decompress(files / "1.0.0.trees.tsz")
ts2 = tskit.load(files / "1.0.0.trees")
assert ts == ts2
def verify(self, ts):
with tempfile.TemporaryDirectory() as tmpdir:
path = pathlib.Path(tmpdir) / "treeseq.tsz"
tszip.compress(ts, path)
other_ts = tszip.decompress(path)
self.assertEqual(ts.tables, other_ts.tables)
