def test_roundtrip(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig = SourmashSignature(e)
s = save_signatures([sig])
siglist = list(load_signatures(s))
sig2 = siglist[0]
e2 = sig2.minhash
assert sig.similarity(sig2) == 1.0
assert sig2.similarity(sig) == 1.0
def test_compare_ne(track_abundance):
# same content, different names -> different
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig1 = SourmashSignature(e, name='foo')
f = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
f.add("AT" * 10)
sig2 = SourmashSignature(f, name='bar')
assert sig1 != sig2
def test_load_one_fail_multisig(track_abundance):
e1 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig1 = SourmashSignature(e1)
e2 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig2 = SourmashSignature(e2)
x = save_signatures([sig1, sig2])
with pytest.raises(ValueError):
y = load_one_signature(x)
def test_compare(track_abundance):
# same content, same name -> equal
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig1 = SourmashSignature(e, name='foo')
f = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
f.add("AT" * 10)
sig2 = SourmashSignature(f, name='foo')
assert e == f
def test_roundtrip(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig = SourmashSignature(e)
s = save_signatures([sig])
siglist = list(load_signatures(s))
sig2 = siglist[0]
e2 = sig2.minhash
assert sig.similarity(sig2) == 1.0
assert sig2.similarity(sig) == 1.0
def test_roundtrip_empty(track_abundance):
# edge case, but: empty minhash? :)
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e)
s = save_signatures([sig])
siglist = list(load_signatures(s))
sig2 = siglist[0]
e2 = sig2.minhash
assert sig.similarity(sig2) == 0
assert sig2.similarity(sig) == 0
def test_memmap():
e1 = sourmash.MinHash(n=1, ksize=20)
sig1 = SourmashSignature(e1)
e2 = sourmash.MinHash(n=1, ksize=25)
sig2 = SourmashSignature(e2)
siglist = [sig1, sig2]
memmapped, filename = to_memmap(np.array(siglist))
# Assert that the data didn't change as a result of memory-mapping
np.testing.assert_array_equal(memmapped, siglist)
assert filename.endswith(".mmap")
def test_compare_ne2_reverse(track_abundance):
# same content, one has filename, other does not -> different
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig1 = SourmashSignature(e, name='foo')
f = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
f.add("AT" * 10)
sig2 = SourmashSignature(f, filename='b')
assert sig2 != sig1
assert sig1 != sig2
def test_roundtrip_empty(track_abundance):
# edge case, but: empty minhash? :)
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e)
s = save_signatures([sig])
siglist = list(load_signatures(s))
sig2 = siglist[0]
e2 = sig2.minhash
assert sig.similarity(sig2) == 0
assert sig2.similarity(sig) == 0
def test_str(track_abundance):
# signatures should be printable
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig = SourmashSignature(e)
print(sig)
assert str(sig) == 'SourmashSignature(59502a74)'
assert repr(sig) == 'SourmashSignature(59502a74)'
sig.d['name'] = 'fizbar'
assert str(sig) == 'SourmashSignature(\'fizbar\', 59502a74)'
assert repr(sig) == 'SourmashSignature(\'fizbar\', 59502a74)'
def test_str(track_abundance):
# signatures should be printable
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig = SourmashSignature(e)
print(sig)
assert str(sig) == 'SourmashSignature(59502a74)'
assert repr(sig) == 'SourmashSignature(59502a74)'
sig._name = 'fizbar'
assert str(sig) == 'SourmashSignature(\'fizbar\', 59502a74)'
assert repr(sig) == 'SourmashSignature(\'fizbar\', 59502a74)'
def test_roundtrip_max_hash(track_abundance):
e = sourmash.MinHash(n=0, ksize=20, track_abundance=track_abundance,
max_hash=10)
e.add_hash(5)
sig = SourmashSignature(e)
s = save_signatures([sig])
siglist = list(load_signatures(s))
sig2 = siglist[0]
e2 = sig2.minhash
assert e.max_hash == e2.max_hash
assert sig.similarity(sig2) == 1.0
assert sig2.similarity(sig) == 1.0
def test_save_minified(track_abundance):
e1 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig1 = SourmashSignature(e1, name="foo")
e2 = sourmash.MinHash(n=1, ksize=25, track_abundance=track_abundance)
sig2 = SourmashSignature(e2, name="bar baz")
x = save_signatures([sig1, sig2])
assert '\n' not in x
assert len(x.split('\n')) == 1
y = list(load_signatures(x))
assert len(y) == 2
assert any(sig.name() == 'foo' for sig in y)
assert any(sig.name() == 'bar baz' for sig in y)
def test_save_load_multisig_json():
e1 = sourmash.MinHash(n=1, ksize=20)
sig1 = SourmashSignature(e1)
e2 = sourmash.MinHash(n=1, ksize=25)
sig2 = SourmashSignature(e2)
x = save_signatures_json([sig1, sig2])
y = list(load_signatures_json(x))
print(x)
assert len(y) == 2
assert sig1 in y # order not guaranteed, note.
assert sig2 in y
assert sig1 != sig2
def test_save_load_multisig(track_abundance):
e1 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig1 = SourmashSignature(e1)
e2 = sourmash.MinHash(n=1, ksize=25, track_abundance=track_abundance)
sig2 = SourmashSignature(e2)
x = save_signatures([sig1, sig2])
y = list(load_signatures(x))
print(x)
assert len(y) == 2
assert sig1 in y # order not guaranteed, note.
assert sig2 in y
assert sig1 != sig2
def test_binary_fp(tmpdir, track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
path = tmpdir.join("1.sig")
with open(str(path), 'wb') as fp:
sig = SourmashSignature(e)
s = save_signatures([sig], fp)
def test_load_one_succeed(track_abundance):
e1 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig1 = SourmashSignature(e1)
x = save_signatures([sig1])
y = load_one_signature(x)
assert sig1 == y
def test_hashable(track_abundance):
# check: can we use signatures as keys in dictionaries and sets?
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add("AT" * 10)
sig = SourmashSignature(e)
x = set()
x.add(sig)
def test_load_compressed(track_abundance):
e1 = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig1 = SourmashSignature(e1)
x = save_signatures([sig1], compression=5)
y = load_one_signature(x)
assert sig1 == y
sigfile = utils.get_test_data('genome-s10+s11.sig.gz')
sigs = load_signatures(sigfile)
def test_similarity_downsample(track_abundance):
e = sourmash.MinHash(n=0,
ksize=20,
track_abundance=track_abundance,
max_hash=2**63)
f = sourmash.MinHash(n=0,
ksize=20,
track_abundance=track_abundance,
max_hash=2**2)
e.add_hash(1)
e.add_hash(5)
assert len(e.get_mins()) == 2
f.add_hash(1)
f.add_hash(5) # should be discarded due to max_hash
assert len(f.get_mins()) == 1
ee = SourmashSignature(e)
ff = SourmashSignature(f)
with pytest.raises(ValueError): # mismatch in max_hash
ee.similarity(ff)
x = ee.similarity(ff, downsample=True)
assert round(x, 1) == 1.0
def test_similarity_downsample(track_abundance):
e = sourmash.MinHash(n=0, ksize=20, track_abundance=track_abundance,
max_hash=2**63)
f = sourmash.MinHash(n=0, ksize=20, track_abundance=track_abundance,
max_hash=2**2)
e.add_hash(1)
e.add_hash(5)
assert len(e.get_mins()) == 2
f.add_hash(1)
f.add_hash(5) # should be discarded due to max_hash
assert len(f.get_mins()) == 1
ee = SourmashSignature(e)
ff = SourmashSignature(f)
with pytest.raises(ValueError): # mismatch in max_hash
ee.similarity(ff)
x = ee.similarity(ff, downsample=True)
assert round(x, 1) == 1.0
def test_md5(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add_hash(5)
sig = SourmashSignature(e)
print(sig._save())
assert sig.md5sum() == 'eae27d77ca20db309e056e3d2dcd7d69', sig.md5sum()
def test_name_3(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e, name='foo',
filename='foo.txt')
assert sig.name() == 'foo'
def test_name_4(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e)
assert sig.name() == sig.md5sum()[:8]
def test_name_4(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e)
assert sig.name() == sig.md5sum()[:8]
def test_name_3(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
sig = SourmashSignature(e, name='foo', filename='foo.txt')
assert sig.name() == 'foo'
def test_md5(track_abundance):
e = sourmash.MinHash(n=1, ksize=20, track_abundance=track_abundance)
e.add_hash(5)
sig = SourmashSignature(e)
assert sig.md5sum() == 'eae27d77ca20db309e056e3d2dcd7d69', sig.md5sum()
def sketch(name, sequence, params={'ksize': 21, 'n': 0, 'scaled': 100}):
mh = MinHash(**params)
mh.add_sequence(sequence, force=True)
# "force" will sketch Ns in DNA sequences as well
sig = SourmashSignature(mh, name=name)
return sig
