def test_fit_null_binary(self):
p = np.loadtxt(P_BINARY)
m = np.loadtxt(M)
cov = pd.DataFrame([])
# no covariates
params = np.array([
-1.41572498, 0.35847998, -0.03014792, 2.46252819, 0.96908425,
-0.20952455, -0.27988125, 0.36798503, -0.03278285, -1.34132024,
0.844149
])
null_res = fit_null(p, m, cov, False, firth=False)
self.assertTrue(abs((params - null_res.params).max()) < 1E-7)
# no covariates, firth regression
null_res = fit_null(p, m, cov, False, firth=True)
self.assertAlmostEqual(null_res, -57.884527394557985)
# covariates
cov = np.loadtxt(COV)
cov = pd.DataFrame(cov)
params = np.array([
-0.87072948, 0.26456701, 0.03485904, 2.80243184, 1.086393,
-0.3882244, -0.46883396, 0.61387846, 0.09962477, -1.45376984,
0.93929299, 0.07927743, -1.54631396, 0.1098796
])
null_res = fit_null(p, m, cov, False, firth=False)
self.assertTrue(abs((params - null_res.params).max()) < 1E-7)
# covariates, firth regression
null_res = fit_null(p, m, cov, False, firth=True)
self.assertAlmostEqual(null_res, -55.60790630835098)
# perfectly separable data
p = np.array([1] * 10 + [0] * 90)
m = np.array([1] * 10 + [0] * 90).reshape(-1, 1)
cov = pd.DataFrame([])
self.assertEqual(fit_null(p, m, cov, False, False), None)
def test_fit_null_continuous(self):
p_cont = np.loadtxt(P_CONT)
m = np.loadtxt(M)
# no covariates
params = np.array([0.65572473, -0.16129649, 0.03417796, -0.08011702,
0.10902641, 0.00599514, -0.09081684, -0.13653787,
0.17798003, -0.16793408, 0.12959982])
null_res = fit_null(p_cont, m, pd.DataFrame([]), True, firth=False)
self.assertTrue(abs((params - null_res.params).max())
< 1E-7)
# covariates
cov = np.loadtxt(COV)
cov = pd.DataFrame(cov)
params = np.array([0.49070237, -0.17284083, 0.00710691, -0.11784811,
0.07352861, 0.01219004, -0.04772721, -0.17089199,
0.18198025, -0.17141095, 0.11330439, 0.08887165,
0.20304982, 0.13802362])
null_res = fit_null(p_cont, m, cov, True, firth=False)
self.assertTrue(abs((params - null_res.params).max())
< 1E-7)
def test_find_enet_selected_binary(self):
p = pd.read_csv(P,
index_col=0,
sep='\t')['binary']
b = np.array([5.60000000e-01, 2.16450216e-37, -1.81966726e-37,
1.99034682e-38, -1.34400134e-37, -2.94000294e-38,
-1.90213827e-37, 8.08080808e-38,
1.89393939e-37])
idx = [10, 29, 39, 95, 110, 153, 156, 164]
g = find_enet_selected(b, idx, p, np.array([[]]), 'vcf',
None, False, None,
VariantFile(VCF), set(p.index),
None, False, False, None, False)
v = next(g)
self.assertEqual(v.kmer, 'FM211187_83_G_A')
self.assertEqual(v.af, 0.28)
self.assertTrue(abs(v.prep - 0.17050715825327736) < 1E-7)
self.assertTrue(np.isnan(v.pvalue))
self.assertTrue(abs(v.kbeta - 2.164502164502162e-37) < 1E-7)
self.assertEqual(v.max_lineage, None)
self.assertEqual(v.kstrains,
[
'sample_10', 'sample_11', 'sample_13', 'sample_18',
'sample_19',
'sample_20', 'sample_23', 'sample_25', 'sample_26',
'sample_31', 'sample_34', 'sample_36', 'sample_40',
'sample_45'
]
)
self.assertEqual(v.nkstrains,
[
'sample_1', 'sample_12', 'sample_14', 'sample_15',
'sample_16', 'sample_17', 'sample_2', 'sample_21',
'sample_22', 'sample_24', 'sample_27', 'sample_28',
'sample_29', 'sample_3', 'sample_30', 'sample_32',
'sample_33', 'sample_35', 'sample_37', 'sample_38',
'sample_39', 'sample_4', 'sample_41', 'sample_42',
'sample_43', 'sample_44', 'sample_46', 'sample_47',
'sample_48', 'sample_49', 'sample_5', 'sample_50',
'sample_6', 'sample_7', 'sample_8', 'sample_9'
]
)
self.assertEqual(len(v.notes), 0)
# read to exhaustion
for v in g:
pass
self.assertEqual(v.kmer, 'FM211187_3592_G_A')
# with fixed effects
pf = np.loadtxt(PFIRTH)
mf = np.loadtxt(MFIRTH)
cov = pd.DataFrame([])
null_res = fit_null(pf, mf, cov, False, firth=False).llr
null_firth = fit_null(pf, mf, cov, False, firth=True)
g = find_enet_selected(b, idx, p, np.array([[]]), 'vcf',
(mf, null_res, null_firth), False,
None, VariantFile(VCF), set(p.index),
None, False, False, None, False)
v = next(g)
self.assertEqual(v.kmer, 'FM211187_83_G_A')
self.assertEqual(v.af, 0.28)
self.assertTrue(abs(v.prep - 0.17050715825327736) < 1E-7)
self.assertEqual(v.pvalue, 1)
self.assertTrue(abs(v.kbeta - 2.164502164502162e-37) < 1E-7)
self.assertEqual(v.max_lineage, None)
self.assertEqual(v.kstrains,
[
'sample_10', 'sample_11', 'sample_13', 'sample_18',
'sample_19',
'sample_20', 'sample_23', 'sample_25', 'sample_26',
'sample_31', 'sample_34', 'sample_36', 'sample_40',
'sample_45'
]
)
self.assertEqual(v.nkstrains,
[
'sample_1', 'sample_12', 'sample_14', 'sample_15',
'sample_16', 'sample_17', 'sample_2', 'sample_21',
'sample_22', 'sample_24', 'sample_27', 'sample_28',
'sample_29', 'sample_3', 'sample_30', 'sample_32',
'sample_33', 'sample_35', 'sample_37', 'sample_38',
'sample_39', 'sample_4', 'sample_41', 'sample_42',
'sample_43', 'sample_44', 'sample_46', 'sample_47',
'sample_48', 'sample_49', 'sample_5', 'sample_50',
'sample_6', 'sample_7', 'sample_8', 'sample_9'
]
)
self.assertEqual(len(v.notes), 0)
# read to exhaustion
for v in g:
pass
self.assertEqual(v.kmer, 'FM211187_3592_G_A')