def test_create_dataset__aa_with_bad_codon(self):
"""Test when trying to translate 'N--' codon. Should translate to X"""
dataset_obj = Dataset.objects.create()
v = Vouchers.objects.get(code="CP100-10")
seq = Sequences.objects.get(code=v, gene_code="COI-begin")
seq.sequences = "TCAN--CGTCCC"
seq.save()
create_dataset(
taxonset_id=1,
geneset_id=1,
gene_codes_ids=[],
voucher_codes='CP100-10',
file_format='FASTA',
outgroup='',
positions='ALL',
partition_by_positions='by gene',
translations=False,
aminoacids=True,
degen_translations='normal',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
dataset_obj.refresh_from_db()
self.assertEqual([], dataset_obj.warnings)
self.assertEqual([], dataset_obj.errors)
def test_create_dataset(self):
"""Test that gaps have not been converted to underscores."""
seq = Sequences.objects.get(code="CP100-10", gene_code="COI-begin")
this_seq = list(seq.sequences)
this_seq[-3:] = '---'
seq.sequences = "".join(this_seq)
seq.save()
dataset_obj = Dataset.objects.create()
create_dataset(
taxonset_id=TaxonSets.objects.last().id,
geneset_id=GeneSets.objects.last().id,
gene_codes_ids=[2], # COI-begin
voucher_codes='CP100-10',
file_format='FASTA',
outgroup='',
positions='ALL',
partition_by_positions='by gene',
translations=False,
aminoacids=False,
degen_translations='NORMAL',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
dataset_obj.refresh_from_db()
self.assertFalse("___" in str(dataset_obj.content))
def test_fasta_with_seqs_of_different_sizes(self):
"""Test that an error message is shown to the users GUI."""
seq = Sequences.objects.get(code="CP100-10", gene_code="COI-begin")
seq.sequences += 'AAAAAAAAAA'
seq.save()
dataset_obj = Dataset.objects.create()
create_dataset(
taxonset_id=TaxonSets.objects.last().id,
geneset_id=GeneSets.objects.last().id,
gene_codes_ids=[2],
voucher_codes='',
file_format='FASTA',
outgroup='',
positions='ALL',
partition_by_positions='by gene',
translations=False,
aminoacids=False,
degen_translations='normal',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
expected = "Matrix Nchar 4749 does not match data length (4739) for taxon"
dataset_obj.refresh_from_db()
self.assertTrue(expected in str(dataset_obj.errors))
def test_create_dataset2(self):
dataset_obj = Dataset.objects.create()
create_dataset(
taxonset_id=1,
geneset_id=1,
gene_codes_ids=[],
voucher_codes='CP100-10',
file_format='FASTA',
outgroup='',
positions='ALL',
partition_by_positions='by gene',
translations=False,
aminoacids=False,
degen_translations='normal',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
expected = ">CP100_10_Aus_aus\nACGACGACGACGACGACGACGACGACGACGACGACGACGACGACGACGACGACGACGACG"
dataset_obj.refresh_from_db()
self.assertIn(expected, dataset_obj.content)
def test_create_dataset_degenerated(self):
dataset_obj = Dataset.objects.create()
gene = Genes.objects.get(gene_code="RpS2")
create_dataset(
taxonset_id=TaxonSets.objects.last().id,
geneset_id=GeneSets.objects.last().id,
gene_codes_ids=[gene.id],
voucher_codes='CP100-10',
file_format='FASTA',
outgroup='',
positions='ALL',
partition_by_positions='by gene',
translations=True,
aminoacids=False,
degen_translations='normal',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
expected = 'MGNMGNMGNMGNMGNMGNMGNMGNMG'
dataset_obj.refresh_from_db()
self.assertTrue(expected in str(dataset_obj.content))
def test_create_dataset_degenerated_warning_data_cannot_be_of_partial_codons(
self):
dataset_obj = Dataset.objects.create()
create_dataset(
taxonset_id=TaxonSets.objects.last().id,
geneset_id=GeneSets.objects.last().id,
gene_codes_ids=[4],
voucher_codes='CP100-10',
file_format='FASTA',
outgroup='',
positions='1st',
partition_by_positions='by gene',
translations=True,
aminoacids=False,
degen_translations='normal',
special=False,
taxon_names=['CODE', 'GENUS', 'SPECIES'],
number_genes='',
introns='YES',
dataset_obj_id=dataset_obj.id,
)
dataset_obj.refresh_from_db()
expected = 'Cannot degenerate codons if you have not selected all codon positions'
self.assertTrue(expected in dataset_obj.errors)