def test_gather_data(self):
'''test gather_data'''
lines = [
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t95\t98.42\tctg_name\t279\t24.4\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\tsome free text',
'ariba_ref1\tref1\t1\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tI14L\t1\tI14L\tSNP\t13\t13\tA\t84\t84\tT\t17\tT\t17\tnon_coding1:0:0:I14L:.:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A\t10,30\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A,G\t20,10,10\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
]
data_dicts = [
summary_cluster.SummaryCluster.line2dict(x) for x in lines
]
cluster = summary_cluster.SummaryCluster()
for data_dict in data_dicts:
cluster.add_data_dict(data_dict)
cluster.gather_data()
expected_summary = {
'assembled': 'yes',
'match': 'yes',
'ref_seq': 'ref1',
'pct_id': '98.33',
'known_var': 'yes',
'novel_var': 'no',
}
self.assertEqual(expected_summary, cluster.summary)
cluster_vars = [
summary_cluster_variant.SummaryClusterVariant(x)
for x in data_dicts
]
expected_variants = {x for x in cluster_vars if x.has_nonsynon}
self.assertEqual(expected_variants, cluster.variants)
def test_init(self):
'''test __init__'''
lines = [
'ariba_ref1\tref1\t1\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tI14L\t1\tI14L\tSNP\t13\t13\tA\t84\t84\tT\t17\tT\t17\tnon_coding1:0:0:I14L:.:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t17\tT\t17\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A\t10,30\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A,G\t20,10,10\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t0\t.\t.\t13\t13\tA\t84\t84\tA\t100\tA,T\t90,10\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
]
expected = [
{
'coding': True,
'known': True,
'var_string': 'I14L',
'var_group': '.',
'het_percent': None
},
{
'coding': False,
'known': True,
'var_string': '14T',
'var_group': 'id1',
'het_percent': 100.0
},
{
'coding': False,
'known': True,
'var_string': '14AT',
'var_group': 'id1',
'het_percent': 25.0
},
{
'coding': False,
'known': True,
'var_string': '14AGT',
'var_group': 'id1',
'het_percent': 50.0
},
{
'coding': False,
'known': True,
'var_string': '14AT',
'var_group': 'id1',
'het_percent': 10.0
},
]
assert len(lines) == len(expected)
for i in range(len(lines)):
data_dict = summary_cluster.SummaryCluster.line2dict(lines[i])
cluster_var = summary_cluster_variant.SummaryClusterVariant(
data_dict)
for key in expected[i]:
got_value = eval('cluster_var.' + key)
self.assertEqual(expected[i][key], got_value)
def _get_all_nonsynon_variants_set(cls, data_dicts):
variants = set()
for data_dict in data_dicts:
cluster_var = summary_cluster_variant.SummaryClusterVariant(
data_dict)
if cluster_var.has_nonsynon or cluster_var.is_het:
variants.add(cluster_var)
return variants
def _has_known_variant(cls, data_dict):
if data_dict['has_known_var'] == '1':
return 'yes'
elif data_dict['known_var'] == '0':
return 'no'
elif data_dict['gene'] == '1': # we don't yet call hets in genes
return 'no'
else:
cluster_var = summary_cluster_variant.SummaryClusterVariant(
data_dict)
return 'het' if cluster_var.is_het else 'no'
def _has_nonsynonymous(cls, data_dict):
cluster_var = summary_cluster_variant.SummaryClusterVariant(data_dict)
has_non_het = data_dict['ref_ctg_effect'] != 'SYN' and \
(
data_dict['has_known_var'] == '1' or \
(data_dict['known_var'] != '1' and (data_dict['ref_ctg_change'] != '.' or data_dict['ref_ctg_effect'] != '.'))
)
if has_non_het and not cluster_var.is_het:
return 'yes'
else:
return 'het' if cluster_var.is_het else 'no'
def test_get_all_nonsynon_variants_set(self):
'''test _get_all_nonsynon_variants_set'''
lines = [
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t95\t98.42\tctg_name\t279\t24.4\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\t.\tsome free text',
'ariba_ref1\tref1\t1\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tI14L\t1\tI14L\tSNP\t13\t13\tA\t84\t84\tT\t17\tT\t17\tnon_coding1:0:0:I14L:.:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A\t10,30\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
'ariba_ref1\tref1\t0\t0\t531\t78\tcluster1\t120\t100\t98.33\tctg_name\t279\t24.4\t1\tSNP\tn\tA14T\t1\tA14T\tSNP\t13\t13\tA\t84\t84\tT\t40\tT,A,G\t20,10,10\tnon_coding1:0:0:A14T:id1:foo_bar\tspam eggs',
]
data_dicts = [
summary_cluster.SummaryCluster.line2dict(x) for x in lines
]
cluster_vars = [
summary_cluster_variant.SummaryClusterVariant(x)
for x in data_dicts
]
expected = {x for x in cluster_vars if x.has_nonsynon}
got = summary_cluster.SummaryCluster._get_all_nonsynon_variants_set(
data_dicts)
self.assertEqual(expected, got)
