def test_max_intron_length_filter(self):
intron_length = MaxIntronLengthFilter(30)
# Create a mock sequence
seq = Sequence()
# Give the sequence some genes
seq.genes = [
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo1'),
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo2'),
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo3')
]
# Give the mock mrnas some exon's
test_mrna0 = Mock()
test_mrna0.identifier = 'foo1-RA'
test_mrna0.death_flagged = False
test_mrna0.exon = Mock()
test_mrna0.get_shortest_intron = Mock(return_value=20)
test_mrna0.get_longest_intron = Mock(return_value=20)
test_mrna1 = Mock()
test_mrna1.identifier = 'foo2-RA'
test_mrna1.death_flagged = False
test_mrna1.exon = Mock()
test_mrna1.get_shortest_intron = Mock(return_value=30)
test_mrna1.get_longest_intron = Mock(return_value=30)
test_mrna2 = Mock()
test_mrna2.identifier = 'foo2-RB'
test_mrna2.death_flagged = False
test_mrna2.exon = None
test_mrna3 = Mock()
test_mrna3.identifier = 'foo3-RA'
test_mrna3.death_flagged = False
test_mrna3.exon = Mock()
test_mrna3.get_shortest_intron = Mock(return_value=40)
test_mrna3.get_longest_intron = Mock(return_value=40)
# Give the mock genes some mrnas
seq.genes[0].mrnas = [test_mrna0]
seq.genes[0].death_flagged = False
seq.genes[1].mrnas = [test_mrna1, test_mrna2]
seq.genes[1].death_flagged = False
seq.genes[2].mrnas = [test_mrna3]
seq.genes[2].death_flagged = False
# Apply the filter
intron_length.apply(seq)
self.assertEqual(len(seq.genes), 3)
self.assertEqual(seq.genes[0].mrnas, [test_mrna0])
self.assertEqual(seq.genes[1].mrnas, [test_mrna1, test_mrna2])
def test_max_cds_length_filter(self):
cds_length = MaxCDSLengthFilter(100)
# Create a mock sequence
seq = Sequence()
# Give the sequence some genes
seq.genes = [
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo1'),
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo2'),
Gene('foo_seq', 'geib_labs', [1, 2], '+', 'foo3')
]
# Give the mock mrnas some cds's
test_mrna0 = Mock()
test_mrna0.identifier = 'foo1-RA'
test_mrna0.death_flagged = False
test_mrna0.cds = Mock()
test_mrna0.cds.length = Mock(return_value=90)
test_mrna1 = Mock()
test_mrna1.identifier = 'foo2-RA'
test_mrna1.death_flagged = False
test_mrna1.cds = None
test_mrna2 = Mock()
test_mrna2.identifier = 'foo2-RB'
test_mrna2.death_flagged = False
test_mrna2.cds = Mock()
test_mrna2.cds.length = Mock(return_value=100)
test_mrna3 = Mock()
test_mrna3.identifier = 'foo3-RA'
test_mrna3.death_flagged = False
test_mrna3.cds = Mock()
test_mrna3.cds.length = Mock(return_value=110)
# Give the mock genes some mrnas
seq.genes[0].mrnas = [test_mrna0]
seq.genes[0].death_flagged = False
seq.genes[1].mrnas = [test_mrna1, test_mrna2]
seq.genes[1].death_flagged = False
seq.genes[2].mrnas = [test_mrna3]
seq.genes[2].death_flagged = False
# Apply the filter
cds_length.apply(seq)
self.assertEqual(len(seq.genes), 3)
self.assertEqual(seq.genes[0].mrnas, [test_mrna0])
self.assertEqual(seq.genes[1].mrnas, [test_mrna1, test_mrna2])
def process_gene_line(self, line):
"""Extracts arguments from a line and instantiates a Gene object."""
kwargs = self.extract_gene_args(line)
if not kwargs:
return
gene_id = kwargs['identifier']
self.genes[gene_id] = Gene(**kwargs)
def test_gene_initialized_without_annotations(self):
newgene = Gene(seq_name="seq1",
source="maker",
indices=[1, 50],
strand="+",
identifier="foo_gene_1")
self.assertFalse(newgene.annotations)
self.assertEquals(0, len(newgene.annotations.keys()))
def __init__(self, population_size, filename_of_genes, max_weight, logger):
self.population_size = population_size
self.genes = Gene(filename_of_genes, logger)
self.max_weight = max_weight
self.max_score_for_run = None
self.logger = logger
self.best_chromosome = np.array([0, 0])
return
def test_gene_initialized_with_annotations(self):
newgene = Gene(seq_name="seq1",
source="maker",
indices=[1, 50],
strand="+",
identifier="foo_gene_1",
annotations={"bar": ["cat"]})
self.assertTrue(newgene.annotations)
self.assertEquals(1, len(newgene.annotations.keys()))
def test_to_tbl_negative(self):
gene = Gene("seq1", "maker", [1, 50], "-", "foo_gene_1")
mrna1 = Mock()
mrna1.to_tbl.return_value = "mrna1_to_tbl...\n"
mrna2 = Mock()
mrna2.to_tbl.return_value = "mrna2_to_tbl...\n"
gene.mrnas.append(mrna1)
gene.mrnas.append(mrna2)
expected = "50\t1\tgene\n\t\t\tlocus_tag\tfoo_gene_1\nmrna1_to_tbl...\nmrna2_to_tbl...\n"
self.assertEquals(gene.to_tbl(), expected)
def process_gene_line(self, line, gene_type):
"""Extracts arguments from a line and instantiates a Gene object."""
kwargs = self.extract_gene_args(line)
if not kwargs:
return
gene_id = kwargs['identifier']
gene = Gene(**kwargs)
if gene_type == 'pseudogene':
gene.pseudo = True
self.genes[gene_id] = gene
def setUp(self):
self.test_gene0 = Gene(seq_name="sctg_0080_0020",
source="maker",
indices=[3734, 7436],
strand='+',
identifier=1)
self.test_gene1 = Gene(seq_name="sctg_0080_0020",
source="maker",
indices=[3734, 7436],
strand='+',
identifier=1)
self.fake_mrna1 = Mock()
self.fake_mrna1.identifier = "fake_mrna1"
self.fake_mrna1.death_flagged = False
self.fake_mrna2 = Mock()
self.fake_mrna2.identifier = "fake_mrna2"
self.fake_mrna2.death_flagged = False
self.test_gene1.mrnas.append(self.fake_mrna1)
self.test_gene1.mrnas.append(self.fake_mrna2)
def test_to_tbl_positive(self):
gene = Gene(seq_name="seq1",
source="maker",
indices=[1, 50],
strand="+",
identifier="foo_gene_1")
self.assertFalse(gene.annotations)
gene.add_annotation('foo', 'dog')
mrna1 = Mock()
mrna1.to_tbl.return_value = "mrna1_to_tbl...\n"
mrna2 = Mock()
mrna2.to_tbl.return_value = "mrna2_to_tbl...\n"
gene.mrnas.append(mrna1)
gene.mrnas.append(mrna2)
expected = "1\t50\tgene\n\t\t\tlocus_tag\tfoo_gene_1\nmrna1_to_tbl...\nmrna2_to_tbl...\n"
self.assertEquals(gene.to_tbl(), expected)