def test_open_mutations_without_header(self):
''' test that we fail for files without correct headers
'''
lines = [['TEST', '1', '200', 'A', 'C', 'missense_variant']]
temp = self.write_temp(lines)
with self.assertRaises(ValueError):
open_mutations(temp.name)
def test_open_mutations_consequence_noninteger_positions(self):
''' test that we fail if the nucleotide position is not an integer
'''
lines = [['symbol', 'chrom', 'pos', 'ref', 'alt', 'consequence'],
['TEST', '1', '200.0', 'A', 'C', 'missense_variant']]
temp = self.write_temp(lines)
with self.assertRaises(ValueError):
open_mutations(temp.name)
def test_open_mutations_multiple_genes(self):
''' test that we load mutations in multiple genes correctly
'''
lines = [['symbol', 'chrom', 'pos', 'ref', 'alt', 'consequence'],
['TEST', '1', '200', 'A', 'C', 'missense_variant'],
['TEST2', '1', '300', 'A', 'C', 'missense_variant']]
temp = self.write_temp(lines)
self.assertEqual(
open_mutations(temp.name), {
'TEST': [{
'chrom': '1',
'pos': 200,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}],
'TEST2': [{
'chrom': '1',
'pos': 300,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}]
})
def main():
args = get_options()
ensembl = EnsemblRequest(args.cache, args.genome_build)
cadd = pysam.TabixFile(args.cadd)
constraint = load_regional_constraint(args.constraint)
# open de novo mutations
all_de_novos = open_mutations(args.de_novos)
mut_dict = load_mutation_rates()
output = open(args.output, 'w')
output.write('symbol\tseverity_p_value\n')
for symbol in all_de_novos:
if symbol in ['', '.']:
continue
print(symbol)
de_novos = all_de_novos[symbol]
p_value = analyse_gene(ensembl, mut_dict, cadd, symbol, de_novos,
constraint, WEIGHTS)
line = '{}\t{}\n'.format(symbol, p_value)
output.write(line)
def test_open_mutations_consequence(self):
''' test that we load indels correctly
'''
lines = [['symbol', 'chrom', 'pos', 'ref', 'alt', 'consequence'],
['TEST', '1', '200', 'A', 'C', 'synonymous_variant']]
temp = self.write_temp(lines)
# check that variants that do not alter protein sequence are excluded
self.assertEqual(open_mutations(temp.name), {})
def test_open_mutations_indels(self):
''' test that we load indels correctly
'''
lines = [['symbol', 'chrom', 'pos', 'ref', 'alt', 'consequence'],
['TEST', '1', '200', 'A', 'C', 'missense_variant'],
['TEST2', '1', '300', 'A', 'CG', 'frameshift_variant']]
temp = self.write_temp(lines)
# check that indels are excluded by default
self.assertEqual(
open_mutations(temp.name), {
'TEST': [{
'chrom': '1',
'pos': 200,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}]
})
# check that if we pass the indels=True argument, then we pick up indels
self.assertEqual(
open_mutations(temp.name, indels=True), {
'TEST': [{
'chrom': '1',
'pos': 200,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}],
'TEST2': [{
'chrom': '1',
'pos': 300,
'ref': 'A',
'alt': 'CG',
'consequence': 'frameshift_variant'
}]
})
def test_open_mutations_column_order(self):
''' test that we can load files with different column orders
'''
# scramble the column order
lines = [['chrom', 'ref', 'alt', 'consequence', 'symbol', 'pos'],
['1', 'A', 'C', 'missense_variant', 'TEST', '200']]
temp = self.write_temp(lines)
# check that we still pick up the correct columns based on the header
self.assertEqual(
open_mutations(temp.name), {
'TEST': [{
'chrom': '1',
'pos': 200,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}]
})
# add an extra column, and check that the file is loaded as per usual
lines = [[
'chrom', 'ref', 'alt', 'consequence', 'symbol', 'pos', 'extra'
], ['1', 'A', 'C', 'missense_variant', 'TEST', '200', 'something']]
self.assertEqual(
open_mutations(temp.name), {
'TEST': [{
'chrom': '1',
'pos': 200,
'ref': 'A',
'alt': 'C',
'consequence': 'missense_variant'
}]
})