From bf500c5814ce24d0b87591daed40b35ee1504bf1 Mon Sep 17 00:00:00 2001 From: nkumar2 Date: Wed, 4 Feb 2026 11:06:17 +0000 Subject: [PATCH] add property for duplicate SS acc QC in accessioning pipeline properties --- ebi_eva_internal_pyutils/spring_properties.py | 6 ++++-- tests/internal/test_spring_prop.py | 15 +++++++++------ 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/ebi_eva_internal_pyutils/spring_properties.py b/ebi_eva_internal_pyutils/spring_properties.py index e7c1c89..25db94c 100644 --- a/ebi_eva_internal_pyutils/spring_properties.py +++ b/ebi_eva_internal_pyutils/spring_properties.py @@ -134,7 +134,8 @@ def _common_accessioning_clustering_properties(self, *, assembly_accession, read def get_accessioning_properties(self, *, read_preference='primary', job_name=None, target_assembly=None, fasta=None, assembly_report=None, project_accession=None, aggregation='BASIC', - taxonomy_accession=None, vcf_file='', output_vcf='', chunk_size=100): + taxonomy_accession=None, vcf_file='', output_vcf='', duplicate_ss_acc_file='', + chunk_size=100): """Properties for accessioning pipeline.""" return self._format( self._common_accessioning_clustering_properties(assembly_accession=target_assembly, @@ -150,7 +151,8 @@ def get_accessioning_properties(self, *, read_preference='primary', job_name=Non 'parameters.taxonomyAccession': taxonomy_accession, 'parameters.vcfAggregation': aggregation, 'parameters.vcf': vcf_file, - 'parameters.outputVcf': output_vcf + 'parameters.outputVcf': output_vcf, + 'parameters.duplicateSSAccFile': duplicate_ss_acc_file }, ) diff --git a/tests/internal/test_spring_prop.py b/tests/internal/test_spring_prop.py index 53230df..0254210 100644 --- a/tests/internal/test_spring_prop.py +++ b/tests/internal/test_spring_prop.py @@ -165,7 +165,8 @@ def test_get_accessioning_properties(self): parameters.taxonomyAccession=9906 parameters.vcfAggregation=BASIC parameters.vcf=/path/to/vcf_file.vcf -parameters.outputVcf= +parameters.outputVcf=/path/to/output_vcf.vcf +parameters.duplicateSSAccFile=/path/to/duplicate_ss_acc_file.vcf accessioning.submitted.categoryId=ss accessioning.clustered.categoryId=rs @@ -181,7 +182,8 @@ def test_get_accessioning_properties(self): assert self.prop.get_accessioning_properties( job_name='SUBSNP_ACCESSION_JOB', target_assembly='GCA_00000001.1', fasta='/path/to/fasta.fa', assembly_report='/path/to/assembly_report.txt', project_accession='PRJEB0001', aggregation='BASIC', - taxonomy_accession='9906', vcf_file='/path/to/vcf_file.vcf') == expected + taxonomy_accession='9906', vcf_file='/path/to/vcf_file.vcf', output_vcf='/path/to/output_vcf.vcf', + duplicate_ss_acc_file='/path/to/duplicate_ss_acc_file.vcf') == expected def test_get_variant_load_properties(self): expected = '''spring.data.mongodb.uri=mongodb://mongouser:mongop%40ssword@mongos-host1.example.com:27017,mongos-host2.example.com:27017/?retryWrites=true&authSource=admin @@ -383,7 +385,8 @@ def test_get_accessioning_properties_with_none(self): parameters.taxonomyAccession=9906 parameters.vcfAggregation=BASIC parameters.vcf=/path/to/vcf_file.vcf -parameters.outputVcf= +parameters.outputVcf=/path/to/output_vcf.vcf +parameters.duplicateSSAccFile=/path/to/duplicate_ss_acc_file.vcf accessioning.submitted.categoryId=ss accessioning.clustered.categoryId=rs @@ -397,9 +400,9 @@ def test_get_accessioning_properties_with_none(self): recovery.cutoff.days=9999999 ''' assert self.prop.get_accessioning_properties( - target_assembly='GCA_00000001.1', fasta=None, - assembly_report=None, project_accession='PRJEB0001', aggregation='BASIC', - taxonomy_accession='9906', vcf_file='/path/to/vcf_file.vcf') == expected + target_assembly='GCA_00000001.1', fasta=None, assembly_report=None, project_accession='PRJEB0001', + aggregation='BASIC', taxonomy_accession='9906', vcf_file='/path/to/vcf_file.vcf', + output_vcf='/path/to/output_vcf.vcf', duplicate_ss_acc_file='/path/to/duplicate_ss_acc_file.vcf') == expected def test_get_release_properties(self): expected = '''spring.data.mongodb.uri=mongodb://mongouser:mongop%40ssword@mongos-host1.example.com:27017,mongos-host2.example.com:27017/?retryWrites=true&authSource=admin