def test_run_kmer_counts_vectors(self, tmpdir):
infasta = "tests/data/example.fa"
infasta = pkg_resources.resource_filename("seekr", infasta)
mean_vector = "tests/data/example_mean.npy"
mean_vector = pkg_resources.resource_filename("seekr", mean_vector)
std_vector = "tests/data/example_std.npy"
std_vector = pkg_resources.resource_filename("seekr", std_vector)
outfile = str(tmpdir.join("2mers_vectors.npy"))
console_scripts._run_kmer_counts(
fasta=infasta,
outfile=outfile,
kmer=2,
binary=True,
centered=False,
standardized=False,
log2=Log2.post.name,
remove_labels=True,
mean_vector=mean_vector,
std_vector=std_vector,
alphabet="AGTC",
)
kmers = np.load(outfile)
expected = "tests/data/example_2mers_count.npy"
expected = pkg_resources.resource_filename("seekr", expected)
expected = np.load(expected)
assert np.allclose(kmers, expected)
def test_run_kmer_counts_raw_csv(self, tmpdir):
infasta = 'tests/data/example.fa'
infasta = pkg_resources.resource_filename('seekr', infasta)
outfile = str(tmpdir.join('3mers.csv'))
console_scripts._run_kmer_counts(fasta=infasta,
outfile=outfile,
kmer=3,
binary=False,
centered=False,
standardized=False,
log2=False,
remove_labels=True,
mean_vector=None,
std_vector=None)
kmers = pd.read_csv(outfile, header=None)
expected = 'tests/data/example_3mers_raw.csv'
expected = pkg_resources.resource_filename('seekr', expected)
expected = pd.read_csv(expected, header=None)
assert np.allclose(kmers.values, expected.values)
def test_run_kmer_counts(self, tmpdir):
infasta = 'tests/data/example.fa'
infasta = pkg_resources.resource_filename('seekr', infasta)
outfile = str(tmpdir.join('2mers.npy'))
console_scripts._run_kmer_counts(fasta=infasta,
outfile=outfile,
kmer=2,
binary=True,
centered=True,
standardized=True,
log2=True,
remove_labels=True,
mean_vector=None,
std_vector=None)
kmers = np.load(outfile)
expected = 'tests/data/example_2mers.npy'
expected = pkg_resources.resource_filename('seekr', expected)
expected = np.load(expected)
assert np.allclose(kmers, expected)
def test_run_kmer_counts_raw_csv(self, tmpdir):
infasta = "tests/data/example.fa"
infasta = pkg_resources.resource_filename("seekr", infasta)
outfile = str(tmpdir.join("3mers.csv"))
console_scripts._run_kmer_counts(
fasta=infasta,
outfile=outfile,
kmer=3,
binary=False,
centered=False,
standardized=False,
log2=Log2.none.name,
remove_labels=True,
mean_vector=None,
std_vector=None,
alphabet="AGTC",
)
kmers = pd.read_csv(outfile, header=None)
expected = "tests/data/example_3mers_raw.csv"
expected = pkg_resources.resource_filename("seekr", expected)
expected = pd.read_csv(expected, header=None)
print(kmers.values)
print(expected)
assert np.allclose(kmers.values, expected.values)