def test_invalid_denom(self, get_mp_data):
num = "Firm"
denom = "beyblade"
with pytest.raises(ValueError) as excinfo:
qarcoal(get_mp_data.table, get_mp_data.taxonomy, num, denom)
denom_err = "No feature(s) found matching denominator string!"
assert denom_err == str(excinfo.value)
def test_both_invalid(self, get_mp_data):
num = "beyblade"
denom = "yugioh"
with pytest.raises(ValueError) as excinfo:
qarcoal(get_mp_data.table, get_mp_data.taxonomy, num, denom)
both_err = ("No feature(s) found matching either numerator or "
"denominator string!")
assert both_err == str(excinfo.value)
def test_shared_features_allowed(self, get_mp_data):
num = "Firmicutes"
denom = "Bacilli"
qarcoal(
get_mp_data.table,
get_mp_data.taxonomy,
num,
denom,
allow_shared_features=True,
)
def test_shared_features_disallowed(self, get_mp_data):
num = "Firmicutes"
denom = "Bacilli"
with pytest.raises(ValueError) as excinfo:
qarcoal(
get_mp_data.table,
get_mp_data.taxonomy,
num,
denom,
allow_shared_features=False,
)
shared_err = "Shared features between num and denom!"
assert shared_err == str(excinfo.value)
def test_negative_counts(self):
samps = ["S{}".format(i) for i in range(3)]
feats = ["S{}".format(i) for i in range(4)]
mat = [np.random.randint(1, 10) for i in range(12)]
mat = np.reshape(mat, (4, 3))
mat[0] = -1
table = biom.table.Table(mat, feats, samps)
tax_labels = ["AB", "BC", "CD", "DE"]
confidence = ["0.99"] * 4
taxonomy = pd.DataFrame([feats, tax_labels, confidence]).T
taxonomy.columns = ["feature-id", "Taxon", "Confidence"]
taxonomy.set_index("feature-id", inplace=True, drop=True)
with pytest.raises(ValueError) as excinfo:
qarcoal(table, taxonomy, "A", "C")
assert "Feature table has negative counts!" == str(excinfo.value)
def test_samples_to_use(self, get_mp_data):
metadata_url = os.path.join(MP_URL, "sample-metadata.tsv")
sample_metadata = pd.read_csv(metadata_url,
sep="\t",
index_col=0,
skiprows=[1],
header=0)
gut_samples = sample_metadata[sample_metadata["BodySite"] == "gut"]
num_gut_samples = gut_samples.shape[0]
gut_samples = Metadata(gut_samples)
num = "p__Bacteroidetes"
denom = "p__Firmicutes"
q = qarcoal(
get_mp_data.table,
get_mp_data.taxonomy,
num,
denom,
samples_to_use=gut_samples,
)
assert q.shape[0] == num_gut_samples
def get_mp_results(get_mp_data):
num = "g__Bacteroides"
denom = "g__Streptococcus"
q = qarcoal(get_mp_data.table, get_mp_data.taxonomy, num, denom)
return q
def test_large_numbers(self):
"""Test large numbers on which Qurro fails.
Qurro fails when |x| > 2^53 - 1 due to JS implementation.
Make a sample feature table with large numbers:
----------------------------------- ** 53
| S0 S1 S2 |
| F0/Charmander 2.0 2.5 2.2 |
| F1/Charmeleon 3.2 2.6 3.5 |
| F2/Charizard 4.1 2.9 3.1 |
| F3/Bulbasaur 6.2 5.2 3.0 |
| F4/Ivysaur 4.3 2.1 2.2 |
| F5/Venusaur 3.7 2.5 4.0 |
-----------------------------------
Num: Char
Denom: saur
Output should be (from WolframAlpha):
-----------------
| log_ratio |
| S0 -21.9188 |
| S1 -30.9458 |
| S2 -7.07556 |
-----------------
"""
samps = ["S{}".format(i) for i in range(3)]
feats = ["F{}".format(i) for i in range(6)]
s0 = [x**53 for x in (2.0, 3.2, 4.1, 6.2, 4.3, 3.7)]
s1 = [x**53 for x in (2.5, 2.6, 2.9, 5.2, 2.1, 2.5)]
s2 = [x**53 for x in (2.2, 3.5, 3.1, 3.0, 2.2, 4.0)]
confidence = ["0.99"] * 6
tax_labels = [
"Charmander",
"Charmeleon",
"Charizard",
"Bulbasaur",
"Ivysaur",
"Vensaur",
]
mat = np.matrix([s0, s1, s2]).T
table = biom.table.Table(mat, feats, samps)
taxonomy = pd.DataFrame([feats, tax_labels, confidence]).T
taxonomy.columns = ["feature-id", "Taxon", "Confidence"]
taxonomy.set_index("feature-id", inplace=True, drop=True)
q = qarcoal(
table,
taxonomy,
"Char",
"saur",
)
wolfram_alpha_vals = np.array([-21.9188, -30.9458, -7.07556])
qarcoal_vals = q.sort_index()["log_ratio"].to_numpy()
diff = wolfram_alpha_vals - qarcoal_vals
# differences are ~ 10^-6
assert diff == pytest.approx(0, abs=1e-5)
