def test_concat_ref(self):
"""Verify the concat ref file is created if and only if requested.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(
directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
concat_ref_file = os.path.join(directory.path, "concat_ref.fasta")
args.concat_ref_file = None
snpmutator.run_from_args(args)
file_exists = os.path.exists(concat_ref_file)
self.assertFalse(
file_exists,
"The concat ref file should not exist when not explicitly requested"
)
args.concat_ref_file = concat_ref_file
snpmutator.run_from_args(args)
file_exists = os.path.exists(concat_ref_file)
self.assertTrue(file_exists,
"The concat ref file is missing when requested.")
def test_not_all_same_pool_mono(self):
"""Verify Mutator creates different mutated fasta files when generating more than one.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(
directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 3
args.num_subs = 2
args.num_insertions = 2
args.num_deletions = 2
args.subset_len = 500
args.mono = True
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
mutated_seq_record3 = read_fasta_seq_record("original_mutated_3.fasta")
self.assertNotEqual(
str(mutated_seq_record1.seq), str(mutated_seq_record2.seq),
"Generated sequences 1 and 2 should be different.")
self.assertNotEqual(
str(mutated_seq_record2.seq), str(mutated_seq_record3.seq),
"Generated sequences 2 and 3 should be different.")
self.assertNotEqual(
str(mutated_seq_record1.seq), str(mutated_seq_record3.seq),
"Generated sequences 1 and 3 should be different.")
def test_summary_creation(self):
"""Verify the summary file is created if and only if requested.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(
directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
summary_file_path = os.path.join(directory.path,
"original_snpListMutated.txt")
args.summary_file = None
snpmutator.run_from_args(args)
summary_file_exists = os.path.exists(summary_file_path)
self.assertFalse(
summary_file_exists,
"The summary file should not exist when not explicitly requested")
args.summary_file = summary_file_path
snpmutator.run_from_args(args)
summary_file_exists = os.path.exists(summary_file_path)
self.assertTrue(summary_file_exists,
"The summary file is missing when requested.")
def test_seqid_override(self):
"""Verify output fasta files can have the defline seqid overridden.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(
directory.path, "original.fasta", 50)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_subs = 3
original_seq_record = read_fasta_seq_record(original_file_path)
args.seq_id = None
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
original_seq_record.id, mutated_seq_record1.id,
"Defline seq id should not change when not requested.")
args.seq_id = "test_override_seqid"
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
mutated_seq_record1.id, args.seq_id,
'Overridden defline seq id "%s" does not match expected value "%s"'
% (mutated_seq_record1.id, args.seq_id))
def test_zero_changes(self):
"""Verify the output fasta file matches the input fasta file when zero substitions, insertions, and deletions are requested.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
snpmutator.run_from_args(args)
no_change = compare_mutated_fasta_files(original_file_path, "original_mutated_1.fasta")
self.assertTrue(no_change, "Generated fasta file does not match original fasta file")
def test_eligible_snp_changes(self):
"""Test substitutions where some positions are ineligible
"""
directory = TempDirectory()
dna = "12345aaaaaAAAAA12345"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_subs = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "12345GGTCTCGTGC12345", "Eligible SNP test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
def test_zero_changes(self):
"""Verify the output fasta file matches the input fasta file when zero substitions, insertions, and deletions are requested.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(
directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
snpmutator.run_from_args(args)
no_change = compare_mutated_fasta_files(original_file_path,
"original_mutated_1.fasta")
self.assertTrue(
no_change,
"Generated fasta file does not match original fasta file")
def test_eligible_snp_changes(self):
"""Test substitutions where some positions are ineligible
"""
directory = TempDirectory()
dna = "12345aaaaaAAAAA12345"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_subs = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "12345GGTCTCGTGC12345",
"Eligible SNP test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
def test_mono_mix_no_del(self):
"""Verify that Monomorphic mutations are the same in all replicates.
"""
directory = TempDirectory()
dna = "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.subset_len = 6
args.mono = True
args.num_sims = 2
args.num_subs = 3
args.num_deletions = 0
args.num_insertions = 3
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(str(mutated_seq_record1.seq), str(mutated_seq_record2.seq), "Monomorphic mix of mutations do not match, mutated seq 1=%s mutated seq 2=%s" % (str(mutated_seq_record1.seq), str(mutated_seq_record2.seq)))
def test_not_all_same(self):
"""Verify Mutator creates different mutated fasta files when generating more than one.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 3
args.num_subs = 2
args.num_insertions = 2
args.num_deletions = 2
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
mutated_seq_record3 = read_fasta_seq_record("original_mutated_3.fasta")
self.assertNotEqual(str(mutated_seq_record1.seq), str(mutated_seq_record2.seq), "Generated sequences 1 and 2 should be different.")
self.assertNotEqual(str(mutated_seq_record2.seq), str(mutated_seq_record3.seq), "Generated sequences 2 and 3 should be different.")
self.assertNotEqual(str(mutated_seq_record1.seq), str(mutated_seq_record3.seq), "Generated sequences 1 and 3 should be different.")
def test_vcf_creation(self):
"""Verify the VCF file is created if and only if requested.
"""
directory = TempDirectory()
original_file_path, dna = write_random_dna_fasta(directory.path, "original.fasta", 1000)
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
vcf_file_path = os.path.join(directory.path, "test.vcf")
args.vcf_file = None
snpmutator.run_from_args(args)
vcf_file_exists = os.path.exists(vcf_file_path)
self.assertFalse(vcf_file_exists, "The VCF file should not exist when not explicitly requested")
args.vcf_file = vcf_file_path
snpmutator.run_from_args(args)
vcf_file_exists = os.path.exists(vcf_file_path)
self.assertTrue(vcf_file_exists, "The VCF file is missing when requested.")
def test_mutate_mix_changes(self):
"""Test a mix of substitutions, inserts, and deletes.
"""
directory = TempDirectory()
dna = "GGGGGGGGGG"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_subs = 1
args.num_insertions = 1
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "GGTGGGGAGG", "Mutate mix 1,1,1 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_subs = 2
args.num_insertions = 2
args.num_deletions = 2
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "GTGTGCGGGC", "Mutate mix 2,2,2 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_subs = 3
args.num_insertions = 4
args.num_deletions = 3
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "GCTGTAGTGAC", "Mutate mix 3,4,3 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
def test_delete_changes(self):
"""Test various numbers of deletions.
"""
directory = TempDirectory()
dna = "GCCAAATCGG"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "GCAAATCGG",
"Delete 1 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_deletions = 5
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "CAACG",
"Delete 5 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_deletions = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "",
"Delete 10 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
def test_insert_changes(self):
"""Test various numbers of insertions.
"""
directory = TempDirectory()
dna = "TTTTAATTTT"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_insertions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "TTTGTAATTTT",
"Insert 1 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_insertions = 5
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "TCTTGTATATATTTC",
"Insert 5 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_insertions = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "TCTATGTTATATTATTTCTC",
"Insert 10 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
def test_mono_mix_no_del(self):
"""Verify that Monomorphic mutations are the same in all replicates.
"""
directory = TempDirectory()
dna = "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.subset_len = 6
args.mono = True
args.num_sims = 2
args.num_subs = 3
args.num_deletions = 0
args.num_insertions = 3
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(
str(mutated_seq_record1.seq), str(mutated_seq_record2.seq),
"Monomorphic mix of mutations do not match, mutated seq 1=%s mutated seq 2=%s"
% (str(mutated_seq_record1.seq), str(mutated_seq_record2.seq)))
def test_pooling(self):
"""Verify that pooling places mutations at the same location in all replicates.
"""
directory = TempDirectory()
dna = "AAAAAAAAAA"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 2
args.subset_len = 1
args.num_subs = 1
args.num_insertions = 0
args.num_deletions = 0
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(
str(mutated_seq_record1.seq), 'AATAAAAAAA',
"Pooling SNP replicate 1 test failed, dna=%s mutated seq1=%s" %
(dna, str(mutated_seq_record1.seq)))
self.assertEqual(
str(mutated_seq_record2.seq), 'AACAAAAAAA',
"Pooling SNP replicate 2 test failed, dna=%s mutated seq2=%s" %
(dna, str(mutated_seq_record2.seq)))
args.num_subs = 0
args.num_insertions = 1
args.num_deletions = 0
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(
str(mutated_seq_record1.seq), 'AAAGAAAAAAA',
"Pooling INS replicate 1 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record1.seq)))
self.assertEqual(
str(mutated_seq_record2.seq), 'AAACAAAAAAA',
"Pooling INS replicate 2 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record2.seq)))
args.num_subs = 0
args.num_insertions = 0
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(
str(mutated_seq_record1.seq), 'AAAAAAAAA',
"Pooling DEL replicate 1 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record1.seq)))
self.assertEqual(
str(mutated_seq_record2.seq), 'AAAAAAAAA',
"Pooling DEL replicate 2 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record1.seq)))
def test_mutate_mix_changes(self):
"""Test a mix of substitutions, inserts, and deletes.
"""
directory = TempDirectory()
dna = "GGGGGGGGGG"
original_file_path = write_fixed_dna_fasta(dna, directory.path,
"original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_subs = 1
args.num_insertions = 1
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "GGTGGGGAGG",
"Mutate mix 1,1,1 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_subs = 2
args.num_insertions = 2
args.num_deletions = 2
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "GTGTGCGGGC",
"Mutate mix 2,2,2 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
args.num_subs = 3
args.num_insertions = 4
args.num_deletions = 3
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(
str(mutated_seq_record.seq), "GCTGTAGTGAC",
"Mutate mix 3,4,3 test failed, dna=%s mutated seq=%s" %
(dna, str(mutated_seq_record.seq)))
def test_pooling(self):
"""Verify that pooling places mutations at the same location in all replicates.
"""
directory = TempDirectory()
dna = "AAAAAAAAAA"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 2
args.subset_len = 1
args.num_subs = 1
args.num_insertions = 0
args.num_deletions = 0
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(str(mutated_seq_record1.seq), 'AATAAAAAAA', "Pooling SNP replicate 1 test failed, dna=%s mutated seq1=%s" % (dna, str(mutated_seq_record1.seq)))
self.assertEqual(str(mutated_seq_record2.seq), 'AACAAAAAAA', "Pooling SNP replicate 2 test failed, dna=%s mutated seq2=%s" % (dna, str(mutated_seq_record2.seq)))
args.num_subs = 0
args.num_insertions = 1
args.num_deletions = 0
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(str(mutated_seq_record1.seq), 'AAAGAAAAAAA', "Pooling INS replicate 1 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record1.seq)))
self.assertEqual(str(mutated_seq_record2.seq), 'AAACAAAAAAA', "Pooling INS replicate 2 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record2.seq)))
args.num_subs = 0
args.num_insertions = 0
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record1 = read_fasta_seq_record("original_mutated_1.fasta")
mutated_seq_record2 = read_fasta_seq_record("original_mutated_2.fasta")
self.assertEqual(str(mutated_seq_record1.seq), 'AAAAAAAAA', "Pooling DEL replicate 1 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record1.seq)))
self.assertEqual(str(mutated_seq_record2.seq), 'AAAAAAAAA', "Pooling DEL replicate 2 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record1.seq)))
def test_insert_changes(self):
"""Test various numbers of insertions.
"""
directory = TempDirectory()
dna = "TTTTAATTTT"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_insertions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "TTTGTAATTTT", "Insert 1 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_insertions = 5
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "TCTTGTATATATTTC", "Insert 5 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_insertions = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "TCTATGTTATATTATTTCTC", "Insert 10 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
def test_delete_changes(self):
"""Test various numbers of deletions.
"""
directory = TempDirectory()
dna = "GCCAAATCGG"
original_file_path = write_fixed_dna_fasta(dna, directory.path, "original.fasta")
args = make_default_args(original_file_path)
args.random_seed = 1
args.num_sims = 1
args.num_deletions = 1
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "GCAAATCGG", "Delete 1 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_deletions = 5
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "CAACG", "Delete 5 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
args.num_deletions = 10
snpmutator.run_from_args(args)
mutated_seq_record = read_fasta_seq_record("original_mutated_1.fasta")
self.assertEqual(str(mutated_seq_record.seq), "", "Delete 10 test failed, dna=%s mutated seq=%s" % (dna, str(mutated_seq_record.seq)))
def _run_snp_mutator(**kwargs):
return run_from_args(_generate_snp_mutator_args(tmp_path, **kwargs))