def test_TRRecord_GetGenotypes():
# Test good example
ref_allele = "CAGCAGCAG"
alt_alleles = ["CAGCAGCAGCAG", "CAGCAGCAGCAGCAGCAG"]
rec = trh.TRRecord(dummy_record1, ref_allele, alt_alleles, "CAG", "", None)
print(rec) # To test str function
true_gts = [[ref_allele, alt_alleles[0]], [alt_alleles[0], alt_alleles[0]],
[alt_alleles[0], alt_alleles[0]],
[alt_alleles[0], alt_alleles[1]],
[alt_alleles[1], alt_alleles[1]], [ref_allele]]
true_len_gts = [[3, 4], [4, 4], [4, 4], [4, 6], [6, 6], [3]]
ind = 0
for sample in rec.vcfrecord:
stringt = rec.GetStringGenotype(sample)
lengt = rec.GetLengthGenotype(sample)
assert (all([(stringt[i] == true_gts[ind][i])
for i in range(len(stringt))]))
assert (all([(lengt[i] == true_len_gts[ind][i])
for i in range(len(lengt))]))
ind += 1
# Test example where alt=[]
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
print(rec) # To test str function
for sample in rec.vcfrecord:
stringt = rec.GetStringGenotype(sample)
lengt = rec.GetLengthGenotype(sample)
assert (all([item == ref_allele for item in stringt]))
assert (all([item == 3 for item in lengt]))
# Test example with discrepancy between alt_alleles and genotypes given
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1, ref_allele, [], "CAG", "", None)
def test_GetMaxAllele():
# Test good example, no samplelist, uselength=True (default)
ref_allele = "CAGCAGCAG"
alt_alleles = ["CAGCAGCAGCAG", "CAGCAGCAGCAGCAGCAG"]
rec = trh.TRRecord(dummy_record1, ref_allele, alt_alleles, "CAG", "", None)
print(rec) # To test str function
true_al_max = 6.0
al_max = rec.GetMaxAllele()
assert al_max == true_al_max
# Test good example with samplelist
true_al_freqs_slist = {ref_allele: 0.4, alt_alleles[0]: 0.6}
true_len_al_freqs_slist = {3: 0.4, 4: 0.6}
slist = ['S1', 'S3', 'S6']
true_al_max_slist = 4.0
al_max_slist = rec.GetMaxAllele(samplelist=slist)
assert al_max_slist == true_al_max_slist
# Test example where alt=[]
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_al_max = 3.0
al_max = rec.GetMaxAllele()
assert al_max == true_al_max
# Test example with non of samples in samplelist in VCF
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_al_max_slist = np.nan
al_max_slist = rec.GetMaxAllele(samplelist=['NonExistentSample'])
assert np.isnan(al_max_slist) == True
# Test example where that has one uncalled sample only
rec = trh.TRRecord(dummy_record4, ref_allele, [], "CAG", "", None)
true_al_max = np.nan
al_max = rec.GetMaxAllele()
assert np.isnan(al_max) == True
def test_TRRecord_full_alleles():
full_ref = "TCAGCAGCAGA"
full_alts = [
"ACAGCAGCAGCAGC",
"ACAGCAGCAGCAGCAGCAGG",
"ACAGCAGCAGCAGG",
"ACAGCAGCAGG",
"TCAGCAGG",
]
ref_allele = full_ref[1:-1]
alt_alleles = []
for allele in full_alts:
alt_alleles.append(allele[1:-1])
motif = 'FOO'
ID = 'BAR'
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1,
None,
None,
motif,
ID,
None,
full_alleles=(full_ref, full_alts))
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1,
ref_allele,
alt_alleles,
motif,
ID,
None,
full_alleles=(["CAGCAGCAQQQQQQQQQQQQQQQ"], full_alts))
with pytest.raises(ValueError):
bad_alts = ["CAGCAGCAQQQQQQQQQQQQQQQ", full_alts[1]]
trh.TRRecord(dummy_record1,
ref_allele,
alt_alleles,
motif,
ID,
None,
full_alleles=(ref_allele, bad_alts))
record = trh.TRRecord(dummy_record1,
ref_allele,
alt_alleles,
motif,
ID,
None,
full_alleles=(ref_allele, alt_alleles))
assert record.UniqueStringGenotypes() == {0, 1, 2, 5}
assert record.UniqueStringGenotypeMapping() == {
0: 0,
1: 1,
2: 2,
3: 1,
4: 0,
5: 5
}
def test_GetAlleleCounts():
# Test good example, no samplelist, uselength=True (default)
ref_allele = "CAGCAGCAG"
alt_alleles = ["CAGCAGCAGCAG", "CAGCAGCAGCAGCAGCAG"]
rec = trh.TRRecord(dummy_record1, ref_allele, alt_alleles, "CAG", "", None)
print(rec) # To test str function
true_al_counts = {ref_allele: 2, alt_alleles[0]: 6, alt_alleles[1]: 3}
true_len_al_counts = {3: 2, 4: 6, 6: 3}
al_counts_uselength = rec.GetAlleleCounts()
al_counts_nolength = rec.GetAlleleCounts(uselength=False)
assert (all(v == true_len_al_counts[k]
for k, v in al_counts_uselength.items())
and len(al_counts_uselength) == len(true_len_al_counts))
assert (all(v == true_al_counts[k] for k, v in al_counts_nolength.items())
and len(al_counts_nolength) == len(true_al_counts))
# Test good example with samplelist
true_al_counts_slist = {ref_allele: 2, alt_alleles[0]: 3}
true_len_al_counts_slist = {3: 2, 4: 3}
slist = ['S1', 'S3', 'S6']
al_counts_uselength_slist = rec.GetAlleleCounts(samplelist=slist)
al_counts_nolength_slist = rec.GetAlleleCounts(samplelist=slist,
uselength=False)
assert (all(v == true_len_al_counts_slist[k]
for k, v in al_counts_uselength_slist.items()) and
len(al_counts_uselength_slist) == len(true_len_al_counts_slist))
assert (all(v == true_al_counts_slist[k]
for k, v in al_counts_nolength_slist.items())
and len(al_counts_nolength_slist) == len(true_al_counts_slist))
# Test example where alt=[]
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_len_al_counts = {3: 9}
al_counts_uselength = rec.GetAlleleCounts()
assert (all(v == true_len_al_counts[k]
for k, v in al_counts_uselength.items())
and len(al_counts_uselength) == len(true_len_al_counts))
# Test example with non of samples in samplelist in VCF
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_len_al_counts_slist = {}
al_counts_uselength_slist = rec.GetAlleleCounts(
samplelist=['NonExistentSample'])
assert (all(v == true_len_al_counts_slist[k]
for k, v in al_counts_uselength_slist.items()) and
len(al_counts_uselength_slist) == len(true_len_al_counts_slist))
# Test example where that has one uncalled sample only
rec = trh.TRRecord(dummy_record4, ref_allele, [], "CAG", "", None)
true_len_al_counts = {}
al_counts_uselength = rec.GetAlleleCounts()
assert (all(v == true_len_al_counts[k]
for k, v in al_counts_uselength.items())
and len(al_counts_uselength) == len(true_len_al_counts))
def test_TRRecord_allele_lengths():
ref_allele = "CAGCAGCAG"
alt_alleles = ["CAGCAGCAGCAG", "CAGCAGCAGCAGCAGCAG"]
motif = 'FOO'
ID = 'BAR'
# alt alleles
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1,
ref_allele,
alt_alleles,
motif,
ID,
"some_field",
alt_allele_lengths=[4, 6])
record = trh.TRRecord(dummy_record1,
ref_allele,
None,
motif,
ID,
"some_field",
alt_allele_lengths=[4, 5.5])
assert record.alt_alleles == [motif * 4, motif * 5 + "F"]
# ref allele
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1,
ref_allele,
alt_alleles,
motif,
ID,
None,
ref_allele_length=5)
with pytest.raises(ValueError):
trh.TRRecord(dummy_record1,
None,
alt_alleles,
motif,
ID,
None,
ref_allele_length=5)
record = trh.TRRecord(dummy_record1,
None,
None,
motif,
ID,
None,
ref_allele_length=5.5,
alt_allele_lengths=[4, 5.5])
assert record.ref_allele == motif * 5 + 'F'
def test_TRRecord_unique_lengths():
record = trh.TRRecord(dummy_record2, "ACGACGACG",
["ACGAAGACG", "ACGACGACGACG", "ACGACGACAACG"], "ACG",
"ACG-repeat", None)
assert record.UniqueLengthGenotypes() == {0, 2}
assert record.UniqueLengthGenotypeMapping() == {0: 0, 1: 0, 2: 2, 3: 2}
def test_TRRecord_print():
ref = "ABC"
alt = ["DEF", "GHI"]
motif = "foo"
ID = "bar"
record = trh.TRRecord(dummy_record1, ref, alt, motif, ID)
assert str(record) == "{} {} {} {},{}".format(ID, motif, ref, alt[0],
alt[1])
record = trh.TRRecord(dummy_record1, ref, alt, motif, None)
assert str(record) == "{}:{} {} {} {},{}".format(dummy_record1.CHROM,
dummy_record1.POS, motif,
ref, alt[0], alt[1])
record = trh.TRRecord(dummy_record1,
"B", ["E", "H"],
motif,
ID,
full_alleles=(ref, alt))
assert str(record) == "{} {} {} {},{}".format(ID, motif, ref, alt[0],
alt[1])
record = trh.TRRecord(dummy_record1,
ref,
None,
motif,
ID,
alt_allele_lengths=[3, 5.5])
assert str(record) == "{} {} {} n_reps:3,n_reps:5.5".format(ID, motif, ref)
record = trh.TRRecord(dummy_record1,
None,
None,
motif,
ID,
ref_allele_length=7,
alt_allele_lengths=[3, 5.5])
assert str(record) == ("{} {} n_reps:7 n_reps:3,n_reps:5.5".format(
ID, motif))
def test_GetGenotypeCounts():
# Test good example, no samplelist, uselength=True (default)
ref_allele = "CAGCAGCAG"
alt_alleles = ["CAGCAGCAGCAG", "CAGCAGCAGCAGCAGCAG"]
rec = trh.TRRecord(dummy_record1, ref_allele, alt_alleles, "CAG", "", None)
print(rec) # To test str function
true_gt_counts = {
(ref_allele, alt_alleles[0]): 1,
(alt_alleles[0], alt_alleles[0]): 2,
(alt_alleles[0], alt_alleles[1]): 1,
(alt_alleles[1], alt_alleles[1]): 1,
(ref_allele, ): 1
}
true_len_gt_counts = {(3, 4): 1, (4, 4): 2, (4, 6): 1, (6, 6): 1, (3, ): 1}
gt_counts_uselength = rec.GetGenotypeCounts()
gt_counts_nolength = rec.GetGenotypeCounts(uselength=False)
assert (all(v == true_len_gt_counts[k]
for k, v in gt_counts_uselength.items())
and len(gt_counts_uselength) == len(true_len_gt_counts))
assert (all(v == true_gt_counts[k] for k, v in gt_counts_nolength.items())
and len(gt_counts_nolength) == len(true_gt_counts))
# Test good example with samplelist
true_gt_counts_slist = {
(ref_allele, alt_alleles[0]): 1,
(alt_alleles[0], alt_alleles[0]): 1,
(ref_allele, ): 1
}
true_len_gt_counts_slist = {(3, 4): 1, (4, 4): 1, (3, ): 1}
slist = ['S1', 'S3', 'S6']
gt_counts_uselength_slist = rec.GetGenotypeCounts(samplelist=slist)
gt_counts_nolength_slist = rec.GetGenotypeCounts(samplelist=slist,
uselength=False)
assert (all(v == true_len_gt_counts_slist[k]
for k, v in gt_counts_uselength_slist.items()) and
len(gt_counts_uselength_slist) == len(true_len_gt_counts_slist))
assert (all(v == true_gt_counts_slist[k]
for k, v in gt_counts_nolength_slist.items())
and len(gt_counts_nolength_slist) == len(true_gt_counts_slist))
# Test example where alt=[]
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_len_gt_counts = {(3, 3, 3): 1, (3, 3): 3}
gt_counts_uselength = rec.GetGenotypeCounts()
assert (all(v == true_len_gt_counts[k]
for k, v in gt_counts_uselength.items())
and len(gt_counts_uselength) == len(true_len_gt_counts))
# Test example with non of samples in samplelist in VCF
rec = trh.TRRecord(dummy_record3, ref_allele, [], "CAG", "", None)
true_len_gt_counts_slist = {}
gt_counts_uselength_slist = \
rec.GetGenotypeCounts(samplelist=['NonExistentSample'])
assert (all(v == true_len_gt_counts_slist[k]
for k, v in gt_counts_uselength_slist.items()) and
len(gt_counts_uselength_slist) == len(true_len_gt_counts_slist))
# Test example where that has one uncalled sample only
rec = trh.TRRecord(dummy_record4, ref_allele, [], "CAG", "", None)
true_len_gt_counts = {}
gt_counts_uselength = rec.GetGenotypeCounts()
assert (all(v == true_len_gt_counts[k]
for k, v in gt_counts_uselength.items())
and len(gt_counts_uselength) == len(true_len_gt_counts))
def test_TRRecord_iter():
record = trh.TRRecord(dummy_record1, "ACG", ["A", "C", "G", "T"], "FOO",
"BAR", "some_field")
record_iter = iter(record)
assert next(record_iter) == dummy_sample1
assert next(record_iter) == dummy_sample2
