def test_region_studied_dict(self):
conv_region_dict = {
"Chromosome": ["X", "X", "M"],
"Start": [50001, 55001, 50001],
"End": [52001, 60601, 60026]
}
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example3.bed')
nocall_filename = os.path.join(os.path.dirname(
__file__), 'NoncallableRegions_example3.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example3_dict.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_example3.json')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(
os.path.dirname(__file__),
'vcf_example3.vcf'),
'GRCh38',
'HG00628',
conv_region_dict=conv_region_dict,
region_studied_filename=region_studied_filename,
nocall_filename=nocall_filename)
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={7: [2, 3], 8: [3, 4]})
def test_region_studied_dict(self):
conv_region_dict = {
"Chromosome": ["X", "X", "M"],
"Start": [50001, 55001, 50001],
"End": [52001, 60601, 60026]
}
self.maxDiff = None
region_studied_filename = os.path.join(os.path.dirname(__file__),'RegionsStudied_example3.bed')
nocall_filename = os.path.join(os.path.dirname(__file__),'NoncallableRegions_example3.bed')
output_filename = os.path.join(os.path.dirname(__file__), self.TEST_RESULT_DIR, 'fhir_example3_dict.json')
expected_outfult_filename = os.path.join(os.path.dirname(__file__),'expected_example3.json')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example3.vcf'), 'GRCh38', 'HG00628', conv_region_dict= conv_region_dict, region_studied_filename=region_studied_filename, nocall_filename=nocall_filename)
bDone = oVcf2Fhir.convert(output_filename)
# check if translation was completed
self.assertEqual(bDone, True)
actual_fhir_json = json.load(open(output_filename))
# Validate the pased sequence relationship
self.assertEqual(_validate_phase_rel(actual_fhir_json, {7: [2, 3], 8 : [3, 4]}), True)
# Validate: list of observation uids and list of result uids same.
# Also set the uids to '' to avoid guid comparison in next step
map_ids = _get_uids_map(actual_fhir_json)
self.assertEqual(map_ids['obv_ids'], map_ids['result_ids'])
actual_fhir_json['issued'] = ''
# Finally, check if the acutal json after removing all uids is same as exppected json
expected_fhir_json = json.load(open(expected_outfult_filename))
self.assertEqual(actual_fhir_json, expected_fhir_json)
def test_region_studied(self):
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example3.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'RegionsToConvert_example3.bed')
nocall_filename = os.path.join(os.path.dirname(
__file__), 'NoncallableRegions_example3.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example3.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_example3.json')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(
os.path.dirname(__file__),
'vcf_example3.vcf'),
'GRCh38',
'HG00628',
conv_region_filename=conv_region_filename,
region_studied_filename=region_studied_filename,
nocall_filename=nocall_filename)
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={7: [2, 3], 8: [3, 4]})
def test_logger_forks(self):
region_studied_filename = os.path.join(os.path.dirname(__file__),
'RegionsStudied_example3.bed')
conv_region_filename = os.path.join(os.path.dirname(__file__),
'RegionsToConvert_example3.bed')
nocall_filename = os.path.join(os.path.dirname(__file__),
'NoncallableRegions_example3.bed')
output_filename = os.path.join(os.path.dirname(__file__), self.LOG_DIR,
'logging_fhir.json')
# create logger
general_logger = logging.getLogger('vcf2fhir.general')
invalid_record_logger = logging.getLogger('vcf2fhir.invalidrecord')
general_logger.setLevel(logging.DEBUG)
invalid_record_logger.setLevel(logging.DEBUG)
# add ch to logger
general_logger.addHandler(self.genearl_fh)
invalid_record_logger.addHandler(self.invalid_record_fh)
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(os.path.dirname(__file__), 'vcf_example3.vcf'),
'GRCh38',
'HG00628',
conv_region_filename=conv_region_filename,
region_studied_filename=region_studied_filename,
nocall_filename=nocall_filename)
o_vcf_2_fhir.convert(output_filename)
self.assertEqual(os.path.exists(self.log_general_filename), True)
self.assertEqual(os.path.exists(self.log_invalid_record_filename),
True)
def test_empty_fhir_json(self):
conv_region_filename = os.path.join(os.path.dirname(__file__),'RegionsToConvert_empty_example4.bed')
output_filename = os.path.join(os.path.dirname(__file__), self.TEST_RESULT_DIR, 'fhir_example4_test.json')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example4.vcf'), 'GRCh38', 'HG00628', conv_region_filename=conv_region_filename)
bDone = oVcf2Fhir.convert(output_filename)
# check if translation was completed
self.assertEqual(bDone, True)
def test_required_ref_build(self):
with self.assertRaises(Exception) as context:
vcf2fhir.Converter(
os.path.join(os.path.dirname(__file__), 'vcf_example1.vcf'))
self.assertEqual(
'You must provide build number ("GRCh37" or "GRCh38")',
str(context.exception))
def test_annotation(self):
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'NB6TK328_region_studied.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'NB6TK328_conversion_region.bed')
annotation_filename = os.path.join(os.path.dirname(
__file__), 'NB6TK328_annotations.txt')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_annotation.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_annotation.json')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(
os.path.dirname(__file__),
'NB6TK328_filtered.vcf'),
'GRCh38',
'NB6TK328',
conv_region_filename=conv_region_filename,
region_studied_filename=region_studied_filename,
ratio_ad_dp=0.95,
genomic_source_class='germline',
annotation_filename=annotation_filename)
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={18: [11, 12], 19: [12, 13]})
def test_tabix(self):
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example4.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'RegionsToConvert_example4.bed')
nocall_filename = os.path.join(os.path.dirname(
__file__), 'NoncallableRegions_example4.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example4_tabix.json')
expected_outfult_filename = os.path.join(
os.path.dirname(__file__), 'expected_example4.json')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example4.vcf.gz'), 'GRCh38', 'HG00628', has_tabix=True,
conv_region_filename=conv_region_filename, region_studied_filename=region_studied_filename, nocall_filename=nocall_filename)
oVcf2Fhir.convert(output_filename)
actual_fhir_json = json.load(open(output_filename))
# Validate the pased sequence relationship
self.assertEqual(_validate_phase_rel(
actual_fhir_json, {31: [24, 25], 32: [25, 26]}), True)
# Validate: list of observation uids and list of result uids same.
# Also set the uids to '' to avoid guid comparison in next step
map_ids = _get_uids_map(actual_fhir_json)
self.assertEqual(map_ids['obv_ids'], map_ids['result_ids'])
actual_fhir_json['issued'] = ''
# Finally, check if the acutal json after removing all uids is same as exppected json
expected_fhir_json = json.load(open(expected_outfult_filename))
self.assertEqual(actual_fhir_json, expected_fhir_json)
def test_tabix(self):
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example4.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'RegionsToConvert_example4.bed')
nocall_filename = os.path.join(os.path.dirname(
__file__), 'NoncallableRegions_example4.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example4_tabix.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_example4.json')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(
os.path.dirname(__file__),
'vcf_example4.vcf.gz'),
'GRCh38',
'HG00628',
has_tabix=True,
conv_region_filename=conv_region_filename,
region_studied_filename=region_studied_filename,
nocall_filename=nocall_filename)
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={31: [24, 25], 32: [25, 26]})
def test_conv_region_region_studied(self):
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example3.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'RegionsToConvert_example3.bed')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example3.vcf'), 'GRCh37',
'abc', conv_region_filename=conv_region_filename, region_studied_filename=region_studied_filename)
self.assertEqual(type(oVcf2Fhir), vcf2fhir.Converter)
def test_empty_fhir_json(self):
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'RegionsToConvert_empty_example4.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example4_test.json')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(
__file__), 'vcf_example4.vcf'), 'GRCh38', 'HG00628', conv_region_filename=conv_region_filename)
oVcf2Fhir.convert(output_filename)
def test_region_studied_only(self):
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'RegionsStudied_example4.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_example4_test.json')
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(
__file__), 'vcf_example4.vcf'), 'GRCh38', 'HG00628', region_studied_filename=region_studied_filename)
oVcf2Fhir.convert(output_filename)
def test_no_conv_region_nocall(self):
nocall_filename = os.path.join(os.path.dirname(
__file__), 'NoncallableRegions_example3.bed')
with self.assertRaises(Exception) as context:
vcf2fhir.Converter(os.path.join(os.path.dirname(
__file__), 'vcf_example3.vcf'), 'GRCh37', 'abc', nocall_filename=nocall_filename)
self.assertEqual(
'Please also provide region_studied_filename when nocall_filename is provided', str(context.exception))
def test_conv_region_dict(self):
conv_region_dict = {
"Chromosome": ["X", "X", "M"],
"Start": [50001, 55001, 50001],
"End": [52001, 60601, 60026]
}
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example3.vcf'), 'GRCh37', 'abc', conv_region_dict=conv_region_dict)
self.assertEqual(type(oVcf2Fhir), vcf2fhir.Converter)
def test_conv_region_only(self):
conv_region_filename = os.path.join(os.path.dirname(__file__),
'RegionsToConvert_example3.bed')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(os.path.dirname(__file__), 'vcf_example3.vcf'),
'GRCh37',
'abc',
conv_region_filename=conv_region_filename)
self.assertEqual(type(o_vcf_2_fhir), vcf2fhir.Converter)
def test_with_patient_id(self):
self.maxDiff = None
o_vcf_2_fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(
__file__), 'vcf_example1.vcf'), 'GRCh37', 'HG00628')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_with_patient_example1.json')
expected_output_filename = os.path.join(os.path.dirname(
__file__), 'expected_example1_with_patient.json')
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={8: [5, 6]})
def test_structural_mixed(self):
self.maxDiff = None
o_vcf_2_fhir = vcf2fhir.Converter(os.path.join(
os.path.dirname(__file__), 'vcf_structural_variants.vcf'),
'GRCh37',
genomic_source_class='mixed',)
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_structural_mixed.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_fhir_mixed_structural.json')
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={49: [38, 40], 48: [36, 38]})
def test_structural_germline(self):
self.maxDiff = None
o_vcf_2_fhir = vcf2fhir.Converter(os.path.join(
os.path.dirname(__file__), 'vcf_structural_variants.vcf'),
'GRCh37',
genomic_source_class='germline',)
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'fhir_structural_germline.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__),
'expected_fhir_germline_structural.json')
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
dict={45: [34, 36], 44: [32, 34]})
def test_with_patient_id(self):
self.maxDiff = None
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example1.vcf'), 'GRCh37', 'HG00628')
output_filename = os.path.join(os.path.dirname(__file__), self.TEST_RESULT_DIR, 'fhir_with_patient_example1.json')
expected_outfult_filename = os.path.join(os.path.dirname(__file__),'expected_example1_with_patient.json')
bDone = oVcf2Fhir.convert(output_filename)
# check if translation was completed
self.assertEqual(bDone, True)
actual_fhir_json = json.load(open(output_filename))
# Validate the pased sequence relationship
self.assertEqual(_validate_phase_rel(actual_fhir_json, {8 : [5, 6]}), True)
# Validate: list of observation uids and list of result uids same.
# Also set the uids to '' to avoid guid comparison in next step
map_ids = _get_uids_map(actual_fhir_json)
self.assertEqual(map_ids['obv_ids'], map_ids['result_ids'])
actual_fhir_json['issued'] = ''
# Finally, check if the acutal json after removing all uids is same as exppected json
expected_fhir_json = json.load(open(expected_outfult_filename))
self.assertEqual(actual_fhir_json, expected_fhir_json)
def test_no_region_examined(self):
self.maxDiff = None
region_studied_filename = os.path.join(
os.path.dirname(__file__), 'HG00403A_studied.bed')
conv_region_filename = os.path.join(os.path.dirname(
__file__), 'HG00403A_convert.bed')
output_filename = os.path.join(os.path.dirname(
__file__), self.TEST_RESULT_DIR, 'HG00403A.json')
expected_output_filename = os.path.join(
os.path.dirname(__file__), 'expected_HG00403A.json')
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(
os.path.dirname(__file__),
'HG00403A.vcf.gz'),
'GRCh37',
has_tabix=True,
conv_region_filename=conv_region_filename,
region_studied_filename=region_studied_filename)
o_vcf_2_fhir.convert(output_filename)
_compare_actual_and_expected_fhir_json(
self,
output_filename=output_filename,
expected_output_filename=expected_output_filename,
is_null=True)
def test_valid_ref_build_38(self):
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(os.path.dirname(__file__), 'vcf_example1.vcf'),
'GRCh38')
self.assertEqual(type(o_vcf_2_fhir), vcf2fhir.Converter)
def test_required_vcf_filename(self):
with self.assertRaises(Exception) as context:
vcf2fhir.Converter()
self.assertTrue(
'You must provide vcf_filename' in str(context.exception))
def test_anotherfile(self):
o_vcf_2_fhir = vcf2fhir.Converter(
os.path.join(os.path.dirname(__file__), 'vcf_example2.vcf'),
'GRCh37', 'HG00628')
o_vcf_2_fhir.convert(output_filename=os.path.join(
os.path.dirname(__file__), self.TEST_RESULT_DIR, 'fhir2.json'))
def test_anotherfile(self):
oVcf2Fhir = vcf2fhir.Converter(os.path.join(os.path.dirname(__file__), 'vcf_example2.vcf'), 'GRCh37', 'HG00628')
bDone = oVcf2Fhir.convert(output_filename=os.path.join(os.path.dirname(__file__), self.TEST_RESULT_DIR,'fhir2.json'))
self.assertEqual(bDone, True)
