def test_should_have_near_zero_RR_genotype_likelihood_for_hom_alt_call(
self):
chr1 = 'chr1'
sample_bank = SampleBank("TTTTTAAAAAAAAAAAAAAAAAAAA", chrom=chr1)
sequence_bank_1 = sample_bank.add_sample_name('sample_1')
sequence_bank_1.add_sequence(".........................",
n_fwd=20,
n_rev=20)
sequence_bank_2 = sample_bank.add_sample_name('sample_2')
sequence_bank_2.add_sequence("............C............",
n_fwd=20,
n_rev=20)
vc_wrapper_builder = VariantCallerBuilderFromSampleBank(
sample_bank, self.work_dir)
variant_output = vc_wrapper_builder.build().run().output_vcf
vcf_expectation = VCFExpectation(self, variant_output)
record_expectation = vcf_expectation.has_record_for_variant(
Variant(chr1, 12, "A", "C"))
sample_expectation = record_expectation.with_sample("sample_1")
sample_expectation.has_genotype("0|0").has_RR_genotype_likelihood(0.0)
def test_should_add_sequences_with_same_reference(self):
sample_bank = SampleBank("AAATTTTGGGGG")
sample_bank.add_sample_name("SAMPLE1")
sample_bank.add_sample_name("SAMPLE2")
self.assertEqual(sample_bank["SAMPLE1"].reference.ref_seq,
sample_bank["SAMPLE2"].reference.ref_seq)
def __run_small_variant_caller(self, refcalls, format):
sample_bank = SampleBank("T")
sample_bank.add_sample_name("TEST").add_sequence(".")
variant_caller_builder = VariantCallerBuilderFromSampleBank(
sample_bank, self.work_dir)
variant_caller_builder.configuration = {} # clear config.
variant_caller = variant_caller_builder.build()
variant_caller.add_additional_command('outputRefCalls', refcalls)
variant_caller.add_additional_command('outputFormat',
"VCF{}".format(format))
variant_caller.run()
with VCFReaderContextManager(variant_caller.output_vcf) as vcf_file:
actual_schema = vcf_file.read_header()
reference = os.path.splitext(
os.path.basename(variant_caller_builder.wecall_input_data.
reference_filename))[0]
expected_schema = wecall_schema(
file_date=datetime.datetime.today().strftime('%F'),
reference=reference,
contigs={
sample_bank.reference.chrom: {
"length": sample_bank.reference.length_minus_deletions()
}
},
add_ref_calls=refcalls,
format=format)
return expected_schema, actual_schema
def test_should_raise_when_adding_existing_sample(self):
sample_bank = SampleBank("AAA")
sample_name = "SAMPLE1"
sample_bank.add_sample_name(sample_name)
self.assertRaisesRegex(
weCallException,
"Sample SAMPLE1 already exists in the SampleBank.",
sample_bank.add_sample_with_seqs_and_quals, sample_name, [])
def test_should_add_sequence_with_quality(self):
sample_bank = SampleBank("AAA")
sample_name = "SAMPLE1"
sample_bank.add_sample_name(sample_name)
sample_bank[sample_name].add_sequence("...", quality_string="007")
read_lists = [
builder.build_reads(0, {}) for builder in sample_bank["SAMPLE1"]
]
reads = [read for read_list in read_lists for read in read_list]
# ascii: "0": "!", "1": "+", "2": "5", "3": "?", "4": "I", "5": "S",
# "6": "]", "7": "g", "8": "q", "9": "{"
self.assertEqual(reads[0].qual, "!!g")
def test_should_return_all_variants(self):
sample_bank = SampleBank("AAATTTTGGGAG")
sample_bank.add_sample_name("SAMPLE1")
sample_bank.add_sample_name("SAMPLE2")
sample_bank["SAMPLE1"].add_sequence(".....G......")
sample_bank["SAMPLE2"].add_sequence("..........*.")
exp_variant1 = Variant(sample_bank.reference.chrom, 5, "T", "G")
exp_variant2 = Variant(sample_bank.reference.chrom, 9, "GA", "G")
self.assertEqual(sample_bank["SAMPLE1"].variants, {exp_variant1})
self.assertEqual(sample_bank["SAMPLE2"].variants, {exp_variant2})
self.assertEqual(sample_bank.variants, {exp_variant1, exp_variant2})
def test_should_place_variants_at_custom_position(self):
sample_bank = SampleBank("AAATTTTGGGAG", 100)
sample_bank.add_sample_name("SAMPLE1")
sample_bank.add_sample_name("SAMPLE2")
sample_bank["SAMPLE1"].add_sequence(".....G......")
sample_bank["SAMPLE2"].add_sequence("..........*.")
exp_variant1 = Variant(sample_bank.reference.chrom, 105, "T", "G")
exp_variant2 = Variant(sample_bank.reference.chrom, 109, "GA", "G")
self.assertEqual(sample_bank["SAMPLE1"].variants, {exp_variant1})
self.assertEqual(sample_bank["SAMPLE2"].variants, {exp_variant2})
self.assertEqual(sample_bank.variants, {exp_variant1, exp_variant2})
def test_can_build_correct_ref_and_bam_file(self):
bank = SampleBank("ATCCT*ATAATAAATAAATAAT")
sample_name = "TEST_SAMPLE"
bank.add_sample_name(sample_name)
bank[sample_name].add_sequence("....CT.........T......")
builder = WecallInputDataBuilder(self.work_dir).with_sample_bank(bank)
input_files = builder.build()
bam_file = pysam.Samfile(input_files.bam_filenames[0], "rb")
for read in bam_file.fetch():
self.assertEqual(read.pos, 0)
self.assertEqual(read.seq, "ATCCCTATAATAAATTAATAAT")
self.assertEqual(read.cigarstring, "5M1I16M")
fasta_file = pysam.Fastafile(input_files.reference_filename)
self.assertEqual(fasta_file.get_reference_length(bank.reference.chrom),
21)
self.assertEqual(fasta_file.fetch(bank.reference.chrom, 0, 21),
"ATCCTATAATAAATAAATAAT")
def test_should_be_able_to_build_bam_and_ref_data_with_multiple_chromosomes(
self):
bank_1 = SampleBank("A" * 10, 0, chrom='10')
bank_1.add_sample_name("sample").add_sequence("." * 10)
bank_2 = SampleBank("T" * 9, 0, chrom='20')
bank_2.add_sample_name("sample").add_sequence("." * 9)
builder = WecallInputDataBuilder(
self.work_dir).with_sample_bank(bank_1).with_sample_bank(bank_2)
input_files = builder.build()
bam_file = pysam.Samfile(input_files.bam_filenames[0], "rb")
for read in bam_file.fetch(reference='20'):
self.assertEqual(read.pos, 0)
self.assertEqual(read.seq, "T" * 9)
self.assertEqual(read.cigarstring, "9M")
for read in bam_file.fetch(reference='10'):
self.assertEqual(read.pos, 0)
self.assertEqual(read.seq, "A" * 10)
self.assertEqual(read.cigarstring, "10M")
print((dir(read)))
def test_can_build_multiple_bam_files(self):
bank = SampleBank("ATCCT*ATAATAAATAAATAAT")
sample_name1 = "TEST_SAMPLE1"
bank.add_sample_name(sample_name1)
bank[sample_name1].add_sequence("....CT.........T......")
sample_name2 = "TEST_SAMPLE2"
bank.add_sample_name(sample_name2)
bank[sample_name2].add_sequence(".....*.G..........*...")
builder = WecallInputDataBuilder(self.work_dir).with_sample_bank(bank)
input_bams = builder.build().bam_filenames
bam_file1 = pysam.Samfile(input_bams[0], "rb")
for read in bam_file1.fetch():
self.assertEqual(read.pos, 0)
self.assertEqual(read.seq, "ATCCCTATAATAAATTAATAAT")
self.assertEqual(read.cigarstring, "5M1I16M")
bam_file2 = pysam.Samfile(input_bams[1], "rb")
for read in bam_file2.fetch():
self.assertEqual(read.pos, 0)
self.assertEqual(read.seq, "ATCCTAGAATAAATAAAAAT")
self.assertEqual(read.cigarstring, "17M1D3M")