def test_PBWT_default_haplotypes(self):
print(
"\nTesting non-templated PBWT matches over default haplotypes...")
haplotypes = ibd.HaplotypeAlignment()
tpbwt = ibd.TPBWTAnalysis(template=[[1]])
ibd_segs = tpbwt.compute_ibd(haplotypes,
L_m=3,
L_f=0,
missing_site_threshold=1,
verbose=False)
true_segs = pd.DataFrame({
'chromosome': ['1'] * 3,
'start': [0, 0, 1],
'end': [3, 5, 4],
'start_cm': [0, 0, 1],
'end_cm': [3, 5, 4],
'start_bp': [0, 0, 1],
'end_bp': [3, 5, 4],
'id1': [0, 1, 2],
'id2': [7, 6, 3],
'id1_haplotype': [0, 0, 0],
'id2_haplotype': [0, 0, 0]
})
#print(ibd_segs)
#print(true_segs)
#self.assertTrue(np.all(ibd_segs.eq(true_segs)))
self.assertTrue(np.array_equal(ibd_segs, true_segs))
def test_1kgp_compress(self):
print(
"\nTesting 1KGP in-sample IBD compute: comparing results from TPBWT-compressed samples and VCF samples..."
)
tpbwt = ibd.TPBWTAnalysis()
uncompressed_ibd_results = pd.DataFrame()
for chromo in ['1', '2', '15', '22']:
vcf_path = TEST_DATA_PATH + '1kgp_vcf/' + chromo + '.vcf'
map_path = TEST_DATA_PATH + '1kgp_vcf/' + chromo + '.map'
haplotypes = ibd.VcfHaplotypeAlignment(vcf_path, map_path)
tpbwt.compress_alignment('compressed_haplotypes_1kgp/',
haplotypes,
verbose=False)
ibd_segs = tpbwt.compute_ibd(haplotypes, L_f=4.0, verbose=False)
uncompressed_ibd_results = uncompressed_ibd_results.append(
ibd_segs)
haplotypes_compressed_1kgp = ibd.CompressedHaplotypeAlignment(
'compressed_haplotypes_1kgp/')
compressed_ibd_results = tpbwt.compute_ibd(haplotypes_compressed_1kgp,
L_f=4.0,
verbose=False)
# we expect 2 IBD segments > 4 cM
self.assertTrue(uncompressed_ibd_results.shape[0] == 2)
self.assertTrue(compressed_ibd_results.shape[0] == 2)
# the segments should be identical
self.assertTrue(
sum(uncompressed_ibd_results['start_cm']) == sum(
uncompressed_ibd_results['start_cm']))
self.assertTrue(
sum(uncompressed_ibd_results['end_cm']) == sum(
uncompressed_ibd_results['end_cm']))
def test_1kgp_vcf(self):
print(
"\nTesting 1KGP in-sample IBD compute: distantly related samples in VCF..."
)
ibd_results = pd.DataFrame()
for chromo in ['1', '2', '15', '22']:
vcf_path = TEST_DATA_PATH + '1kgp_vcf/' + chromo + '.vcf'
map_path = TEST_DATA_PATH + '1kgp_vcf/' + chromo + '.map'
haplotypes = ibd.VcfHaplotypeAlignment(vcf_path, map_path)
tpbwt = ibd.TPBWTAnalysis()
ibd_segs = tpbwt.compute_ibd(haplotypes,
L_f=4.0,
use_phase_correction=False,
verbose=False)
ibd_results = ibd_results.append(ibd_segs)
# we expect 2 IBD segments > 4 cM
self.assertTrue(ibd_results.shape[0] == 2)
# the segment on chromo 1 is self IBD
id1 = ibd_results[ibd_results['chromosome'] == '1']['id1'].iloc[0]
id2 = ibd_results[ibd_results['chromosome'] == '1']['id2'].iloc[0]
self.assertTrue(id1 == id2)
# the other segment should be between on chromo 15 shared between individual 3 and 7
id1 = ibd_results[ibd_results['chromosome'] == '15']['id1'].iloc[0]
id2 = ibd_results[ibd_results['chromosome'] == '15']['id2'].iloc[0]
estimated_pair = set([id1, id2])
true_pair = set([3, 7])
self.assertTrue(estimated_pair == true_pair)
def test_vcf_no_map(self):
print("\nTesting IBD compute over VCF file with out genetic map...")
haplotypes = ibd.VcfHaplotypeAlignment(TEST_DATA_PATH + 'test.vcf')
tpbwt = ibd.TPBWTAnalysis(template=[[1]])
ibd_segs = tpbwt.compute_ibd(haplotypes,
L_m=5,
L_f=3.0,
use_phase_correction=False,
verbose=False)
#print(ibd_segs)
self.assertTrue(ibd_segs.shape[0] == 2)
def test_pedigree_simulated_vcf(self):
print(
"\nTesting simulated in-sample IBD compute: closely related samples in VCF simulated from 1KGP 4 generation pedigree..."
)
est_ibd = pd.DataFrame()
chromosomes = [str(x) for x in range(1, 23)]
for chromo in chromosomes:
vcf_path = TEST_DATA_PATH + 'pedigree_vcf/' + chromo + '.vcf'
map_path = TEST_DATA_PATH + 'pedigree_vcf/maps/' + chromo + '.map'
haplotypes = ibd.VcfHaplotypeAlignment(vcf_path, map_path)
tpbwt = ibd.TPBWTAnalysis()
ibd_segs = tpbwt.compute_ibd(haplotypes,
use_phase_correction=False,
verbose=False)
est_ibd = est_ibd.append(ibd_segs)
est_ibd['length'] = est_ibd['end_bp'] - est_ibd['start_bp']
true_ibd = pd.read_csv(TEST_DATA_PATH + 'pedigree_vcf/true_ibd.csv')
true_ibd = true_ibd[true_ibd['chromosome'] != 'X']
# compute error in total IBD (in bp) and num segments per pair
pairwise_error_ibd = []
pairwise_error_num_seg = []
for id1 in range(11):
for id2 in range(id1 + 1, 12):
m1 = (true_ibd['id1'] == id1) & (true_ibd['id2'] == id2)
m2 = (est_ibd['id1'] == id1) & (est_ibd['id2'] == id2)
s_true = sum(true_ibd[m1]['length'])
s_est = sum(est_ibd[m2]['length'])
c_true = true_ibd[m1].shape[0]
c_est = est_ibd[m2].shape[0]
if s_true != 0:
pairwise_error_ibd.append((s_est - s_true) / float(s_true))
else:
pairwise_error_ibd.append(s_est - s_true)
if c_true != 0:
pairwise_error_num_seg.append(
(c_est - c_true) / float(c_true))
else:
pairwise_error_num_seg.append(c_est - c_true)
# mean pairwise error in total IBD should be < 0.001
self.assertTrue(abs(np.mean(pairwise_error_ibd)) < 0.001)
# mean pairwise error in num IBD segments should be < 0.07
self.assertTrue(abs(np.mean(pairwise_error_num_seg)) < 0.07)
