Make validation to pass for CNA long and study_es_0_inc data

cBioPortal · May 28, 2024 · cc80e56 · cc80e56
1 parent 0bf6bf2
commit cc80e56
Show file tree

Hide file tree

Showing 4 changed files with 2 additions and 12 deletions.
diff --git a/scripts/importer/validateData.py b/scripts/importer/validateData.py
@@ -4718,7 +4718,7 @@ def process_metadata_files(directory, portal_instance, logger, relaxed_mode, str
             if stable_id in stable_ids:
                 # stable id already used in other meta file, give error:
                 logger.error(
-                    'stable_id repeated. It should be unique across all files in a study',
+                    'stable_id repeated. It should be unique across all files in a directory',
                     extra={'filename_': filename,
                            'cause': stable_id})
             else:

diff --git a/tests/system_tests_import_data.py b/tests/system_tests_import_data.py
@@ -109,8 +109,6 @@ def test_incremental_load(self, run_java, locate_jar):
             '--meta', f'{data_directory}/meta_mutations_extended.txt', '--loadMode', 'bulkload', '--update-info', 'False', '--data', f'{data_directory}/data_mutations_extended.maf', '--noprogress')
         cna_discrete_call = call(*common_part, 'org.mskcc.cbio.portal.scripts.ImportProfileData', '--overwrite-existing',
             '--meta', f'{data_directory}/meta_cna_discrete.txt', '--loadMode', 'bulkload', '--update-info', 'False', '--data', f'{data_directory}/data_cna_discrete.txt', '--noprogress')
-        cna_discrete_long_call = call(*common_part, 'org.mskcc.cbio.portal.scripts.ImportProfileData', '--overwrite-existing',
-            '--meta', f'{data_directory}/meta_cna_discrete_long.txt', '--loadMode', 'bulkload', '--update-info', 'False', '--data', f'{data_directory}/data_cna_discrete_long.txt', '--noprogress')
         cna_log2_call = call(*common_part, 'org.mskcc.cbio.portal.scripts.ImportProfileData', '--overwrite-existing',
             '--meta', f'{data_directory}/meta_cna_log2.txt', '--loadMode', 'bulkload', '--update-info', 'False', '--data', f'{data_directory}/data_cna_log2.txt', '--noprogress')
         expression_median_call = call(*common_part, 'org.mskcc.cbio.portal.scripts.ImportProfileData', '--overwrite-existing',
@@ -130,7 +128,6 @@ def test_incremental_load(self, run_java, locate_jar):
             clinical_sample_call,
             mutation_call,
             cna_discrete_call,
-            cna_discrete_long_call,
             cna_log2_call,
             expression_median_call,
             methylation_hm27_call,

diff --git a/...study_es_0_inc/data_cna_discrete_long.txt → ...0_inc/cna_long/data_cna_discrete_long.txt b/...study_es_0_inc/data_cna_discrete_long.txt → ...0_inc/cna_long/data_cna_discrete_long.txt
@@ -5,7 +5,6 @@ AKT3	10000	TCGA-AO-A129-01	-2
 AKT1	207	TCGA-C8-A12K-01	-1				
 AKT1	207	TCGA-BH-NON-EXIST	2				
 AKT1	207	TCGA-AO-A129-01	2				
-# All after the pipe has to be removed							
 AKT2|TEST	208	TCGA-C8-A12K-01	-2				
 AKT2|TEST	208	TCGA-BH-NON-EXIST	2				
 AKT2|TEST	208	TCGA-AO-A129-01	-1	Putative_Driver	Test driver	Class 1	Class annotation
@@ -15,13 +14,9 @@ HRAS	3265	TCGA-AO-A129-01	0
 KRAS	3845	TCGA-C8-A12K-01	0			Class 2	Class annotation
 KRAS	3845	TCGA-BH-NON-EXIST	-2				
 KRAS	3845	TCGA-AO-A129-01	2	Putative_Passenger	Test passenger	Class 2	Class annotation
-# This gene absent in this file, but it's still part of the profile and has to be updated							
-#ATM	472						
-# This line missing the hugo symbol and the gene has to be detected by entrez id							
 	4893	TCGA-C8-A12K-01	-2				
 	4893	TCGA-BH-NON-EXIST	-2				
 	4893	TCGA-AO-A129-01	-1				
-# This line missing the entrez id and the gene has to be detected by hugo symbol							
 BRCA1		TCGA-C8-A12K-01	2				
 BRCA1		TCGA-BH-NON-EXIST	2				
 BRCA1		TCGA-AO-A129-01	0				
@@ -31,7 +26,6 @@ BRAF	673	TCGA-AO-A129-01	-2
 BRCA2	675	TCGA-C8-A12K-01	-1.5				
 BRCA2	675	TCGA-BH-NON-EXIST	2				
 BRCA2	675	TCGA-AO-A129-01	0				
-# This gene is new! the empty values should be set for the already existing samples in the database							
 CDK1	983	TCGA-C8-A12K-01	-2	Putative_Driver			
 CDK1	983	TCGA-BH-NON-EXIST	-2				
 CDK1	983	TCGA-AO-A129-01	2	Putative_Passenger	Test passenger		
diff --git a/...study_es_0_inc/meta_cna_discrete_long.txt → ...0_inc/cna_long/meta_cna_discrete_long.txt b/...study_es_0_inc/meta_cna_discrete_long.txt → ...0_inc/cna_long/meta_cna_discrete_long.txt
@@ -1,9 +1,8 @@
 cancer_study_identifier: study_es_0
 genetic_alteration_type: COPY_NUMBER_ALTERATION
-datatype: DISCRETE
+datatype: DISCRETE_LONG
 stable_id: gistic
 show_profile_in_analysis_tab: true
 profile_description: Putative copy-number from GISTIC 2.0. Values: -2 = homozygous deletion; -1 = hemizygous deletion; 0 = neutral / no change; 1 = gain; 2 = high level amplification.
 profile_name: Putative copy-number alterations from GISTIC
 data_filename: data_cna_discrete_long.txt
-namespaces: CustomNamespace