Merge branch 'demo-rfc80-poc' into demo-rfc80-poc-generic-assay-patie…

…nt-filtering

src/test/java/org/cbioportal/persistence/mybatisclickhouse/GenomicDataFilterTest.java

@@ -86,8 +86,8 @@ public void getMutationCounts() {
         Map<String, Integer> actualMutationCounts = studyViewMapper.getMutationCounts(StudyViewFilterHelper.build(studyViewFilter, null, null), genomicDataFilterMutation);
         Map<String, Integer> expectedMutationCounts = new HashMap<>();
         expectedMutationCounts.put("mutatedCount", 2);
-        expectedMutationCounts.put("notMutatedCount", 2);
-        expectedMutationCounts.put("notProfiledCount", 11);
+        expectedMutationCounts.put("notMutatedCount", 8);
+        expectedMutationCounts.put("notProfiledCount", 5);
         assertThat(actualMutationCounts)
             .usingRecursiveComparison()
             .ignoringCollectionOrder()
@@ -103,7 +103,7 @@ public void getMutationCountsByType() {
         List<GenomicDataCountItem> actualMutationCountsByType = studyViewMapper.getMutationCountsByType(StudyViewFilterHelper.build(studyViewFilter, null, null), List.of(genomicDataFilterMutation));
         List<GenomicDataCountItem> expectedMutationCountsByType = List.of(
             new GenomicDataCountItem("AKT1", "mutations", List.of(
-                new GenomicDataCount("nonsense mutation", "nonsense_mutation", 1, 1),
+                new GenomicDataCount("nonsense mutation", "nonsense_mutation", 2, 1),
                 new GenomicDataCount("missense mutation", "missense_mutation", 1, 1)
             )));
         assertThat(actualMutationCountsByType)

src/test/java/org/cbioportal/persistence/mybatisclickhouse/MolecularProfileCountTest.java

@@ -48,7 +48,7 @@ public void getMolecularProfileCounts() {
 
         var size = molecularProfileCounts.stream().filter(gc->gc.getValue().equals("mutations"))
             .findFirst().get().getCount().intValue();
-        assertEquals(10, size);
+        assertEquals(11, size);
 
     }
 
@@ -66,7 +66,7 @@ public void getMolecularProfileCountsMultipleStudies() {
 
         var size = molecularProfileCounts.stream().filter(gc->gc.getValue().equals("mutations"))
             .findFirst().get().getCount().intValue();
-        assertEquals(10, size);
+        assertEquals(11, size);
 
     }
 
@@ -84,7 +84,7 @@ public void getMolecularProfileCountsMultipleProfilesUnion() {
 
         var sizeMutations = molecularProfileCounts.stream().filter(gc->gc.getValue().equals("mutations"))
             .findFirst().get().getCount().intValue();
-        assertEquals(10, sizeMutations);
+        assertEquals(11, sizeMutations);
 
         var sizeMrna = molecularProfileCounts.stream().filter(gc->gc.getValue().equals("mrna"))
             .findFirst().get().getCount().intValue();
@@ -107,7 +107,7 @@ public void getMolecularProfileCountsMultipleProfilesIntersect() {
 
         var sizeMutations = molecularProfileCounts.stream().filter(gc->gc.getValue().equals("mutations"))
             .findFirst().get().getCount().intValue();
-        assertEquals(9, sizeMutations);
+        assertEquals(10, sizeMutations);
 
 
 

src/test/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapperTest.java

@@ -48,6 +48,7 @@ public class StudyViewMapperTest extends AbstractTestcontainers {
 
     private static final String STUDY_TCGA_PUB = "study_tcga_pub";
     private static final String STUDY_ACC_TCGA = "acc_tcga";
+    private static final String STUDY_GENIE_PUB = "study_genie_pub";
 
     @Autowired
     private StudyViewMapper studyViewMapper;
@@ -130,14 +131,110 @@ public void getTotalProfiledCountsByGene() {
         StudyViewFilter studyViewFilter = new StudyViewFilter();
         studyViewFilter.setStudyIds(List.of(STUDY_TCGA_PUB));
 
-        var totalProfiledCountsMap = studyViewMapper.getTotalProfiledCounts(StudyViewFilterHelper.build(studyViewFilter, null, null),
+        // Testing profiled counts on samples with gene panel data and WES for one study
+        var totalProfiledCountsForMutationsMap = studyViewMapper.getTotalProfiledCounts(StudyViewFilterHelper.build(studyViewFilter, null, null),
             "MUTATION_EXTENDED", List.of());
-
-        assertEquals(3, totalProfiledCountsMap.size());
-
-        var akt2TotalProfiledCounts = totalProfiledCountsMap.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT2")).findFirst();
-        assertTrue(akt2TotalProfiledCounts.isPresent());
-        assertEquals(4, akt2TotalProfiledCounts.get().getNumberOfProfiledCases().intValue());
+        var totalProfiledCountsForCnaMap = studyViewMapper.getTotalProfiledCounts(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "COPY_NUMBER_ALTERATION", List.of());
+        var sampleProfiledCountsForMutationsWithoutPanelDataMap = studyViewMapper.getSampleProfileCountWithoutPanelData(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "MUTATION_EXTENDED");
+        var sampleProfiledCountsForCnaWithoutPanelDataMap = studyViewMapper.getSampleProfileCountWithoutPanelData(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "COPY_NUMBER_ALTERATION");
+
+        // Assert the count of genes with profiled cases for mutations
+        assertEquals(5, totalProfiledCountsForMutationsMap.size());
+        // Assert the count of genes with profiled cases for CNA
+        assertEquals(5, totalProfiledCountsForCnaMap.size());
+        // Assert the profiled counts for mutations without panel data (WES)
+        assertEquals(6, sampleProfiledCountsForMutationsWithoutPanelDataMap);
+        // Assert the profiled counts for CNA without panel data (WES)
+        assertEquals(11, sampleProfiledCountsForCnaWithoutPanelDataMap);
+
+        // Assert the profiled counts for AKT2 mutations
+        // AKT2 is on testpanel2 in STUDY_TCGA_PUB
+        var akt2TotalProfiledCountsForMutations = totalProfiledCountsForMutationsMap.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT2")).findFirst();
+        assertTrue(akt2TotalProfiledCountsForMutations.isPresent());
+        assertEquals(4, akt2TotalProfiledCountsForMutations.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for BRCA1 mutations
+        // BRCA1 is on testpanel1 in STUDY_TCGA_PUB
+        var brca1TotalProfiledCountsForMutations = totalProfiledCountsForMutationsMap.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA1")).findFirst();
+        assertTrue(brca1TotalProfiledCountsForMutations.isPresent());
+        assertEquals(1, brca1TotalProfiledCountsForMutations.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for AKT1 mutations
+        // AKT1 is on both testpanel1 and testpanel2 in STUDY_TCGA_PUB
+        var akt1TotalProfiledCountsForMutations = totalProfiledCountsForMutationsMap.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT1")).findFirst();
+        assertTrue(akt1TotalProfiledCountsForMutations.isPresent());
+        assertEquals(5, akt1TotalProfiledCountsForMutations.get().getNumberOfProfiledCases().intValue());
+
+        // Assert the profiled counts for AKT2 CNA
+        // AKT2 is on testpanel2 in STUDY_TCGA_PUB
+        var akt2TotalProfiledCountsForCna = totalProfiledCountsForCnaMap.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT2")).findFirst();
+        assertTrue(akt2TotalProfiledCountsForCna.isPresent());
+        assertEquals(6, akt2TotalProfiledCountsForCna.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for BRCA1 CNA
+        // BRCA1 is on testpanel1 in STUDY_TCGA_PUB
+        var brca1TotalProfiledCountsForCna = totalProfiledCountsForCnaMap.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA1")).findFirst();
+        assertTrue(brca1TotalProfiledCountsForCna.isPresent());
+        assertEquals(2, brca1TotalProfiledCountsForCna.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for AKT1 CNA
+        // AKT1 is on both testpanel1 and testpanel2 in STUDY_TCGA_PUB
+        var akt1TotalProfiledCountsForCna = totalProfiledCountsForCnaMap.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT1")).findFirst();
+        assertTrue(akt1TotalProfiledCountsForCna.isPresent());
+        assertEquals(8, akt1TotalProfiledCountsForCna.get().getNumberOfProfiledCases().intValue());
+
+        // Testing profiled counts on combined studies
+        studyViewFilter.setStudyIds(List.of(STUDY_TCGA_PUB, STUDY_GENIE_PUB));
+
+        // Testing profiled counts on samples with gene panel data and WES for a combined study
+        var totalProfiledCountsForMutationsMap1 = studyViewMapper.getTotalProfiledCounts(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "MUTATION_EXTENDED", List.of());
+        var totalProfiledCountsForCnaMap1 = studyViewMapper.getTotalProfiledCounts(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "COPY_NUMBER_ALTERATION", List.of());
+        var sampleProfiledCountsForMutationsWithoutPanelDataMap1 = studyViewMapper.getSampleProfileCountWithoutPanelData(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "MUTATION_EXTENDED");
+        var sampleProfiledCountsForCnaWithoutPanelDataMap1 = studyViewMapper.getSampleProfileCountWithoutPanelData(StudyViewFilterHelper.build(studyViewFilter, null, null),
+            "COPY_NUMBER_ALTERATION");
+
+        // Assert the count of genes with profiled cases for mutations in a combined study
+        assertEquals(8, totalProfiledCountsForMutationsMap1.size());
+        // Assert the count of genes with profiled cases for CNA in a combined study
+        assertEquals(8, totalProfiledCountsForCnaMap1.size());
+        // Assert the profiled counts for mutations without panel data (WES) in a combined study
+        assertEquals(8, sampleProfiledCountsForMutationsWithoutPanelDataMap1);
+        // Assert the profiled counts for CNA without panel data (WES) in a combined study
+        assertEquals(12, sampleProfiledCountsForCnaWithoutPanelDataMap1);
+
+        // Assert the profiled counts for BRCA1 mutations
+        // BRCA1 is on testpanel1 in STUDY_TCGA_PUB
+        var brca1TotalProfiledCountsForMutations1 = totalProfiledCountsForMutationsMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA1")).findFirst();
+        assertTrue(brca1TotalProfiledCountsForMutations1.isPresent());
+        assertEquals(1, brca1TotalProfiledCountsForMutations1.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for BRCA2 mutations
+        // BRCA2 is on testpanel3 and testpanel4 in STUDY_GENIE_PUB
+        var brca2TotalProfiledCountsForMutations1 = totalProfiledCountsForMutationsMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA2")).findFirst();
+        assertTrue(brca2TotalProfiledCountsForMutations1.isPresent());
+        assertEquals(2, brca2TotalProfiledCountsForMutations1.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for AKT2 mutations
+        // AKT2 is on testpanel2 in STUDY_TCGA_PUB and testpanel4 in STUDY_GENIE_PUB
+        var akt2TotalProfiledCountsForMutations1 = totalProfiledCountsForMutationsMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT2")).findFirst();
+        assertTrue(akt2TotalProfiledCountsForMutations1.isPresent());
+        assertEquals(4, akt2TotalProfiledCountsForMutations1.get().getNumberOfProfiledCases().intValue());
+
+        // Assert the profiled counts for BRCA1 CNA
+        // BRCA1 is on testpanel1 in STUDY_TCGA_PUB
+        var brca1TotalProfiledCountsForCna1 = totalProfiledCountsForCnaMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA1")).findFirst();
+        assertTrue(brca1TotalProfiledCountsForCna1.isPresent());
+        assertEquals(2, brca1TotalProfiledCountsForCna1.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for BRCA2 CNA
+        // BRCA2 is on testpanel3 and testpanel4 in STUDY_GENIE_PUB
+        var brca2TotalProfiledCountsForCna1 = totalProfiledCountsForCnaMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("BRCA2")).findFirst();
+        assertTrue(brca2TotalProfiledCountsForCna1.isPresent());
+        assertEquals(3, brca2TotalProfiledCountsForCna1.get().getNumberOfProfiledCases().intValue());
+        // Assert the profiled counts for AKT2 CNA
+        // AKT2 is on testpanel2 in STUDY_TCGA_PUB and testpanel4 in STUDY_GENIE_PUB
+        var akt2TotalProfiledCountsForCna1 = totalProfiledCountsForCnaMap1.stream().filter(c -> c.getHugoGeneSymbol().equals("AKT2")).findFirst();
+        assertTrue(akt2TotalProfiledCountsForCna1.isPresent());
+        assertEquals(7, akt2TotalProfiledCountsForCna1.get().getNumberOfProfiledCases().intValue());
     }
 
     @Test

src/test/resources/clickhouse_cgds.sql

@@ -236,7 +236,7 @@ CREATE TABLE genetic_profile_samples
 CREATE TABLE gene_panel
 (
     internal_id Int32,
-    stable_id     String,
+    stable_id     Nullable(String),
     name          Nullable(String),
     description Nullable(String)
 ) ENGINE = MergeTree() ORDER BY internal_id;
@@ -251,7 +251,7 @@ CREATE TABLE gene_panel_list
 -- --------------------------------------------------------
 CREATE TABLE sample_profile
 (
-    panel_id  Int32,
+    panel_id  Nullable(Int32),
     sample_id          Int32,
     genetic_profile_id Int32
 ) ENGINE = MergeTree() ORDER BY sample_id;

src/test/resources/clickhouse_data.sql

@@ -81,6 +81,8 @@ insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,geneti
 insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,genetic_alteration_type,datatype,name,description,show_profile_in_analysis_tab,generic_assay_type) values (12,'study_tcga_pub_mutational_signature',1,'generic_assay','limit-value','mutational_signature values','mutational_signature values',1,'mutational_signature');
 insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,genetic_alteration_type,datatype,name,description,show_profile_in_analysis_tab) values (14,'study_tcga_pub_cna',1,'COPY_NUMBER_ALTERATION','discrete','Copy-number alterations','Copy number alterations (amplifications and deletions) from targeted sequencing.',1);
 insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,genetic_alteration_type,datatype,name,description,show_profile_in_analysis_tab) values (15,'study_tcga_pub_rppa',1,'PROTEIN_LEVEL','log2-value','Protein expression (RPPA)','Protein expression measured by reverse-phase protein array',0);
+insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,genetic_alteration_type,datatype,name,description,show_profile_in_analysis_tab) values (16,'genie_public_mutations',3,'MUTATION_EXTENDED','maf','Somatic mutations','Somatic mutations from targeted sequencing.',1);
+insert into genetic_profile (genetic_profile_id,stable_id,cancer_study_id,genetic_alteration_type,datatype,name,description,show_profile_in_analysis_tab) values (17,'genie_public_cna',3,'COPY_NUMBER_ALTERATION','discrete','Copy-number alterations','Copy number alterations (amplifications and deletions) from targeted sequencing.',1);
 
 insert into genetic_profile_samples (genetic_profile_id,ordered_sample_list) values (2,'1,2,3,4,5,6,7,8,9,10,11,12,13,14,');
 insert into genetic_profile_samples (genetic_profile_id,ordered_sample_list) values (3,'2,3,6,8,9,10,12,13,');
@@ -207,18 +209,28 @@ insert into mutation (mutation_event_id,genetic_profile_id,sample_id,entrez_gene
 
 insert into gene_panel (internal_id,stable_id,description) values (1,'testpanel1','a test panel consisting of a few genes');
 insert into gene_panel (internal_id,stable_id,description) values (2,'testpanel2','another test panel consisting of a few genes');
+insert into gene_panel (internal_id,stable_id,description) values (3,'testpanel3','third test panel consisting of a few genes');
+insert into gene_panel (internal_id,stable_id,description) values (4,'testpanel4','fourth test panel consisting of a few genes');
 
 insert into gene_panel_list (internal_id,gene_id) values (1,207);
 insert into gene_panel_list (internal_id,gene_id) values (1,369);
 insert into gene_panel_list (internal_id,gene_id) values (1,672);
 insert into gene_panel_list (internal_id,gene_id) values (2,207);
 insert into gene_panel_list (internal_id,gene_id) values (2,208);
 insert into gene_panel_list (internal_id,gene_id) values (2,4893);
+insert into gene_panel_list (internal_id,gene_id) values (3,472);
+insert into gene_panel_list (internal_id,gene_id) values (3,673);
+insert into gene_panel_list (internal_id,gene_id) values (3,675);
+insert into gene_panel_list (internal_id,gene_id) values (4,208);
+insert into gene_panel_list (internal_id,gene_id) values (4,673);
+insert into gene_panel_list (internal_id,gene_id) values (4,675);
 
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (1,2,1);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (1,3,1);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (1,4,null);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (1,6,2);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (1,6,1);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (2,2,1);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (2,2,2);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (2,3,1);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (2,4,2);
@@ -263,6 +275,14 @@ insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (13,4,
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (13,6,null);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (14,2,null);
 insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (14,4,null);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (301,16,3);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (301,17,3);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (302,16,3);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (303,17,3);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (304,17,4);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (305,16,null);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (306,16,null);
+insert into sample_profile (sample_id,genetic_profile_id,panel_id) values (306,17,null);
 
 insert into sample_list (list_id,stable_id,category,cancer_study_id,name,description) values (1,'study_tcga_pub_all','other',1,'all tumors','all tumor samples');
 insert into sample_list (list_id,stable_id,category,cancer_study_id,name,description) values (2,'study_tcga_pub_acgh','other',1,'tumors acgh','all tumors with acgh data');