def run(self):
self.warning('Analysing file: %s' % self.param_required('filepath'))
VcfFilterO = GATK(vcf=self.param_required('filepath'),
caller=self.param_required('caller'),
gatk_folder=self.param_required('gatk_folder'),
reference=self.param_required('reference'),
bgzip_folder=self.param('bgzip_folder'),
tabix_folder=self.param('tabix_folder'),
tmp_dir=self.param('tmp_dir'))
ts_filter_level = None
if self.param_is_defined('ts_filter_level'):
ts_filter_level = self.param('ts_filter_level')
outprefix = None
filename = os.path.split(self.param_required('filepath'))[1]
work_dir = None
if self.param_is_defined('work_dir'):
work_dir = self.param_required('work_dir')
else:
work_dir = os.path.split(self.param_required('filepath'))[0]
outprefix = "{0}/{1}".format(work_dir, filename)
threads = 1
if self.param_is_defined('threads'):
threads = self.param('threads')
log_file = None
if self.param_is_defined('log_file'):
log_file = "{0}_{1}.log".format(self.param('log_file'),
time.strftime("%Y%m%d_%H%M%S"))
outfile = VcfFilterO.run_applyrecalibration(
mode=self.param_required('mode'),
recal_file=self.param_required('recal_file'),
ts_filter_level=ts_filter_level,
tranches_file=self.param_required('tranches_file'),
num_threads=threads,
outprefix=outprefix,
log_file=log_file)
self.param('vcf_filt', outfile)
self.param('vcf_filt_ix', outfile + ".tbi")
def vcf_object():
'''Returns an object'''
vcf_file = pytest.config.getoption("--vcf")
gatk_folder = pytest.config.getoption("--gatk_folder")
reference = pytest.config.getoption("--reference")
vcf_object = GATK(vcf=vcf_file,
gatk_folder=gatk_folder,
reference=reference)
return vcf_object
def test_applyrecalibration_tmpdir(clean_tmp):
'''
Test method to run ApplyRecalibration by setting the tmp_dir for Java.
This test throw an error because the 'recal_file' and 'tranches_file' files are fictitious
'''
vcf_file = pytest.config.getoption("--vcf")
gatk_folder = pytest.config.getoption("--gatk_folder")
reference = pytest.config.getoption("--reference")
vcf_object = GATK(vcf=vcf_file,
gatk_folder=gatk_folder,
reference=reference,
tmp_dir='data/tmp')
with pytest.raises(Exception):
vcf_object.run_applyrecalibration(mode='SNP',
recal_file='data/test.recal',
tranches_file='data/test.tranches',
outprefix='data/outdir/test',
compress=False,
verbose=True)
def run(self):
self.warning('Analysing file: %s' % self.param_required('filepath'))
vcf = GATK(vcf=self.param_required('filepath'),
caller=self.param_required('caller'),
gatk_folder=self.param_required('gatk_folder'),
reference=self.param_required('reference'))
optional_params = {}
if self.param_is_defined('annotations'):
optional_params['annotations'] = self.param('annotations')
if self.param_is_defined('intervals'):
optional_params['intervals'] = self.param('intervals')
if self.param_is_defined('max_gaussians'):
optional_params['max_gaussians'] = self.param('max_gaussians')
if self.param_is_defined('tranches'):
optional_params['tranches'] = self.param('tranches')
verbose = False
if self.param_is_defined('verbose'):
verbose = True
log_file = None
if self.param_is_defined('log_file'):
log_file = "{0}_{1}.log".format(self.param('log_file'),
time.strftime("%Y%m%d_%H%M%S"))
d_out = vcf.run_variantrecalibrator(
self.param_required('resources'),
mode=self.param_required('mode'),
outprefix=self.param_required('filepath'),
verbose=verbose,
log_file=log_file,
**optional_params)
self.param('recal_f', d_out['recal_f'])
self.param('tranches_f', d_out['tranches_f'])