add involved cancer studies into StudyViewFilterHelper

cBioPortal · Dec 11, 2024 · c0902d2 · c0902d2
1 parent af84d12
commit c0902d2
Show file tree

Hide file tree

Showing 22 changed files with 127 additions and 101 deletions.
diff --git a/src/main/java/org/cbioportal/model/StudyViewFilterContext.java b/src/main/java/org/cbioportal/model/StudyViewFilterContext.java
@@ -5,7 +5,10 @@
 
 import java.util.List;
 
-public record StudyViewFilterContext( StudyViewFilter studyViewFilter,
-                                     List<CustomSampleIdentifier> customDataFilterSamples) {
+public record StudyViewFilterContext(
+    StudyViewFilter studyViewFilter,
+    List<CustomSampleIdentifier> customDataFilterSamples,
+    List<String> involvedCancerStudies
+) {
 
 }
diff --git a/src/main/java/org/cbioportal/persistence/helper/StudyViewFilterHelper.java b/src/main/java/org/cbioportal/persistence/helper/StudyViewFilterHelper.java
@@ -23,7 +23,8 @@
 public final class StudyViewFilterHelper {
     public static StudyViewFilterHelper build(@Nullable StudyViewFilter studyViewFilter,
                                               @Nullable Map<DataSource, List<MolecularProfile>> genericAssayProfilesMap,
-                                              @Nullable List<CustomSampleIdentifier> customDataSamples) {
+                                              @Nullable List<CustomSampleIdentifier> customDataSamples,
+                                              @Nullable List<String> involvedCancerStudies) {
         if (Objects.isNull(studyViewFilter)) {
             studyViewFilter = new StudyViewFilter();
         }
@@ -33,6 +34,9 @@ public static StudyViewFilterHelper build(@Nullable StudyViewFilter studyViewFil
         if (Objects.isNull(customDataSamples)) {
             customDataSamples = new ArrayList<>();
         }
+        if (Objects.isNull(involvedCancerStudies)) {
+            involvedCancerStudies = new ArrayList<>();
+        }
         if (studyViewFilter.getGenomicDataFilters() != null && !studyViewFilter.getGenomicDataFilters().isEmpty()) {
             List<GenomicDataFilter> mergedGenomicDataFilters = mergeDataFilters(studyViewFilter.getGenomicDataFilters());
             studyViewFilter.setGenomicDataFilters(mergedGenomicDataFilters);
@@ -45,19 +49,22 @@ public static StudyViewFilterHelper build(@Nullable StudyViewFilter studyViewFil
             List<GenericAssayDataFilter> mergedGenericAssayDataFilters = mergeDataFilters(studyViewFilter.getGenericAssayDataFilters());
             studyViewFilter.setGenericAssayDataFilters(mergedGenericAssayDataFilters);
         }
-        return new StudyViewFilterHelper(studyViewFilter, genericAssayProfilesMap, customDataSamples);
+        return new StudyViewFilterHelper(studyViewFilter, genericAssayProfilesMap, customDataSamples, involvedCancerStudies);
     }
 
     private final StudyViewFilter studyViewFilter;
     private final CategorizedGenericAssayDataCountFilter categorizedGenericAssayDataCountFilter;
     private final List<CustomSampleIdentifier> customDataSamples;
+    private final List<String> involvedCancerStudies;
 
     private StudyViewFilterHelper(@NonNull StudyViewFilter studyViewFilter, 
                                   @NonNull Map<DataSource, List<MolecularProfile>> genericAssayProfilesMap,
-                                  @NonNull List<CustomSampleIdentifier> customDataSamples) {
+                                  @NonNull List<CustomSampleIdentifier> customDataSamples,
+                                  @NonNull List<String> involvedCancerStudies) {
         this.studyViewFilter = studyViewFilter;
         this.categorizedGenericAssayDataCountFilter = extractGenericAssayDataCountFilters(studyViewFilter, genericAssayProfilesMap);
         this.customDataSamples = customDataSamples;
+        this.involvedCancerStudies = involvedCancerStudies;
     }
 
     public StudyViewFilter studyViewFilter() {
@@ -71,6 +78,10 @@ public CategorizedGenericAssayDataCountFilter categorizedGenericAssayDataCountFi
     public List<CustomSampleIdentifier> customDataSamples() {
         return this.customDataSamples;
     }
+
+    public List<String> involvedCancerStudies() {
+        return involvedCancerStudies;
+    }
 
     private CategorizedGenericAssayDataCountFilter extractGenericAssayDataCountFilters(final StudyViewFilter studyViewFilter, Map<DataSource, List<MolecularProfile>> genericAssayProfilesMap) {
         if ((studyViewFilter.getGenericAssayDataFilters() == null || genericAssayProfilesMap.isEmpty()))

diff --git a/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMyBatisRepository.java b/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMyBatisRepository.java
@@ -97,7 +97,12 @@ public List<GenomicDataCount> getMolecularProfileSampleCounts(StudyViewFilterCon
     }
 
     public StudyViewFilterHelper createStudyViewFilterHelper(StudyViewFilterContext studyViewFilterContext) {
-        return StudyViewFilterHelper.build(studyViewFilterContext.studyViewFilter() , getGenericAssayProfilesMap(), studyViewFilterContext.customDataFilterSamples());    
+        return StudyViewFilterHelper.build(
+            studyViewFilterContext.studyViewFilter(),
+            getGenericAssayProfilesMap(),
+            studyViewFilterContext.customDataFilterSamples(),
+            studyViewFilterContext.involvedCancerStudies()
+        );    
     }
 
     @Override

diff --git a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
@@ -161,16 +161,7 @@ public List<ClinicalDataCountItem> getClinicalDataCounts(StudyViewFilter studyVi
 
         var context = createContext(studyViewFilter);
 
-        List<String> involvedCancerStudies = new ArrayList<>();
-        if (studyViewFilter.getStudyIds() != null && !studyViewFilter.getStudyIds().isEmpty()) {
-            involvedCancerStudies = studyViewFilter.getStudyIds();
-        } else if (studyViewFilter.getSampleIdentifiers() != null && !studyViewFilter.getSampleIdentifiers().isEmpty()) {
-            Set<String> studyIdSet = new HashSet<>();
-            for (SampleIdentifier sampleIdentifier : studyViewFilter.getSampleIdentifiers()) {
-                studyIdSet.add(sampleIdentifier.getStudyId());
-            }
-            involvedCancerStudies = studyIdSet.stream().toList();
-        }
+        List<String> involvedCancerStudies = context.involvedCancerStudies();
 
         var result = studyViewRepository.getClinicalDataCounts(context, filteredAttributes);
 
@@ -288,7 +279,8 @@ public List<GenomicDataCountItem> getMutationTypeCountsByGeneSpecific(StudyViewF
 
     private StudyViewFilterContext createContext(StudyViewFilter studyViewFilter) {
         List<CustomSampleIdentifier> customSampleIdentifiers = customDataFilterUtil.extractCustomDataSamples(studyViewFilter);
-        return new StudyViewFilterContext(studyViewFilter, customSampleIdentifiers);
+        List<String> involvedCancerStudies = customDataFilterUtil.extractInvolvedCancerStudies(studyViewFilter);
+        return new StudyViewFilterContext(studyViewFilter, customSampleIdentifiers, involvedCancerStudies);
     }
 
     private List<ClinicalDataCountItem> generateDataCountItemsFromDataCounts(List<ClinicalDataCount> dataCounts) {

diff --git a/src/main/java/org/cbioportal/web/columnar/util/CustomDataFilterUtil.java b/src/main/java/org/cbioportal/web/columnar/util/CustomDataFilterUtil.java
@@ -16,10 +16,7 @@
 import org.springframework.stereotype.Component;
 
 import java.math.BigDecimal;
-import java.util.ArrayList;
-import java.util.List;
-import java.util.Map;
-import java.util.Objects;
+import java.util.*;
 import java.util.function.Function;
 import java.util.stream.Collectors;
 import java.util.stream.Stream;
@@ -37,6 +34,24 @@ public CustomDataFilterUtil(StudyViewFilterUtil studyViewFilterUtil, CustomDataS
         this.patientService = patientService;
     }
 
+    public List<String> extractInvolvedCancerStudies(final StudyViewFilter studyViewFilter) {
+        List<String> involvedCancerStudies;
+
+        if (studyViewFilter.getStudyIds() != null && !studyViewFilter.getStudyIds().isEmpty()) {
+            involvedCancerStudies = studyViewFilter.getStudyIds();
+        } else if (studyViewFilter.getSampleIdentifiers() != null && !studyViewFilter.getSampleIdentifiers().isEmpty()) {
+            Set<String> studyIdSet = new HashSet<>();
+            for (SampleIdentifier sampleIdentifier : studyViewFilter.getSampleIdentifiers()) {
+                studyIdSet.add(sampleIdentifier.getStudyId());
+            }
+            involvedCancerStudies = studyIdSet.stream().toList();
+        }
+        else {
+            involvedCancerStudies = Collections.emptyList();
+        }
+
+        return involvedCancerStudies;
+    }
     public List<CustomSampleIdentifier> extractCustomDataSamples(final StudyViewFilter studyViewFilter) {
         if (studyViewFilter == null) {
             return null;

diff --git a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewFilterMapper.xml b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewFilterMapper.xml
@@ -4,12 +4,12 @@
 <mapper  namespace="org.cbioportal.persistence.mybatisclickhouse.StudyViewMapper">
     <sql id="sampleUniqueIdsFromStudyViewFilter">
         <trim prefixOverrides="INTERSECT">
-            <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+            <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                 INTERSECT
                 SELECT sample_unique_id
                 FROM sample_derived
                 WHERE cancer_study_identifier IN
-                <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                     #{studyId}
                 </foreach>
             </if>
@@ -42,14 +42,14 @@
                     JOIN genetic_profile gp ON sample_profile.genetic_profile_id = gp.genetic_profile_id
                     JOIN cancer_study cs ON gp.cancer_study_id = cs.cancer_study_id
                     JOIN sample_derived on sample_profile.sample_id = sample_derived.internal_id
-                    <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+                    <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                         <where>
                             sample_derived.cancer_study_identifier IN
-                            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                                 #{studyId}
                             </foreach>
                             AND
-                            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" close=")" separator="OR">
+                            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" close=")" separator="OR">
                                 <foreach item="genomicProfileId" collection="ANDGroup" open="(" separator=" OR " close=")">
                                     gp.stable_id=concat(#{studyId}, '_', #{genomicProfileId})
                                 </foreach>
@@ -432,9 +432,9 @@
         SELECT sample_unique_id, patient_unique_id, attribute_value
         FROM clinical_data_derived
         WHERE attribute_name = #{clinicalDataFilter.attributeId} AND type='${type}'
-        <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+        <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
             AND cancer_study_identifier IN
-            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                 #{studyId}
             </foreach>
         </if>
@@ -515,9 +515,9 @@
         FROM genetic_alteration_derived
         WHERE profile_type = #{genomicDataFilter.profileType}
             AND hugo_gene_symbol = #{genomicDataFilter.hugoGeneSymbol}
-        <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+        <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
             AND cancer_study_identifier IN
-            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                 #{studyId}
             </foreach>
         </if>
@@ -647,8 +647,8 @@
                 SELECT sample_unique_id
                 FROM sample_derived
                 WHERE cancer_study_identifier IN
-                <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
-                    <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
+                    <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                         #{studyId}
                     </foreach>
                 </if>
@@ -658,9 +658,9 @@
                 FROM sample_to_gene_panel_derived sgp
                     JOIN gene_panel_to_gene_derived gpg ON sgp.gene_panel_id = gpg.gene_panel_id
                 WHERE
-                <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+                <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                     cancer_study_identifier IN
-                    <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                    <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                         #{studyId}
                     </foreach>
                   AND
@@ -672,9 +672,9 @@
                 SELECT DISTINCT sample_unique_id
                 FROM genomic_event_derived
                 WHERE
-                <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+                <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                     cancer_study_identifier IN
-                    <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                    <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                         #{studyId}
                     </foreach>
                   AND
@@ -722,9 +722,9 @@
             FROM genetic_alteration_derived
             WHERE profile_type = #{genomicDataFilter.profileType}
                 AND hugo_gene_symbol = #{genomicDataFilter.hugoGeneSymbol}
-            <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+            <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                 AND cancer_study_identifier IN
-                <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                     #{studyId}
                 </foreach>
             </if>
@@ -734,9 +734,9 @@
         FROM sample_derived sd
             LEFT JOIN cna_query ON sd.sample_unique_id = cna_query.sampleUniqueId
         WHERE
-        <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+        <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
             cancer_study_identifier IN
-            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                 #{studyId}
             </foreach>
             AND

diff --git a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
@@ -348,9 +348,9 @@
                         FROM genetic_profile
                         WHERE
                             patient_level = 1
-                        <if test="studyViewFilterHelper.studyViewFilter.studyIds != null and !studyViewFilterHelper.studyViewFilter.studyIds.isEmpty()">
+                        <if test="studyViewFilterHelper.involvedCancerStudies != null and !studyViewFilterHelper.involvedCancerStudies.isEmpty()">
                             AND stable_id IN
-                            <foreach item="studyId" collection="studyViewFilterHelper.studyViewFilter.studyIds" open="(" separator="," close=")">
+                            <foreach item="studyId" collection="studyViewFilterHelper.involvedCancerStudies" open="(" separator="," close=")">
                                 concat(#{studyId}, '_', #{profileType})
                             </foreach>
                         </if>

diff --git a/src/test/java/org/cbioportal/persistence/mybatisclickhouse/CNAGenesTest.java b/src/test/java/org/cbioportal/persistence/mybatisclickhouse/CNAGenesTest.java
@@ -40,7 +40,7 @@ public class CNAGenesTest extends AbstractTestcontainers {
     public void getCnaGenes() {
         StudyViewFilter studyViewFilter = new StudyViewFilter();
         studyViewFilter.setStudyIds(List.of(STUDY_TCGA_PUB));
-        var alterationCountByGenes = studyViewMapper.getCnaGenes(StudyViewFilterHelper.build(studyViewFilter, null, null),
+        var alterationCountByGenes = studyViewMapper.getCnaGenes(StudyViewFilterHelper.build(studyViewFilter, null, null, null),
             AlterationFilterHelper.build(studyViewFilter.getAlterationFilter()));
         assertEquals(3, alterationCountByGenes.size());
 
@@ -63,7 +63,7 @@ public void getCnaGenesWithAlterationFilter() {
         cnaEventTypeFilterMap.put(CNA.AMP, true);
         alterationFilter.setCopyNumberAlterationEventTypes(cnaEventTypeFilterMap);
 
-        var alterationCountByGenes = studyViewMapper.getCnaGenes(StudyViewFilterHelper.build(studyViewFilter, null, null),
+        var alterationCountByGenes = studyViewMapper.getCnaGenes(StudyViewFilterHelper.build(studyViewFilter, null, null, studyViewFilter.getStudyIds()),
             AlterationFilterHelper.build(alterationFilter));
         assertEquals(2, alterationCountByGenes.size());
 

diff --git a/src/test/java/org/cbioportal/persistence/mybatisclickhouse/ClinicalEventTypeCountsTest.java b/src/test/java/org/cbioportal/persistence/mybatisclickhouse/ClinicalEventTypeCountsTest.java
@@ -39,7 +39,7 @@ public void getClinicalEventTypeCounts() {
         StudyViewFilter studyViewFilter = new StudyViewFilter();
         studyViewFilter.setStudyIds(List.of(STUDY_TCGA_PUB));
 
-        var clinicalEventTypeCounts = studyViewMapper.getClinicalEventTypeCounts(StudyViewFilterHelper.build(studyViewFilter, null, null));
+        var clinicalEventTypeCounts = studyViewMapper.getClinicalEventTypeCounts(StudyViewFilterHelper.build(studyViewFilter, null, null, studyViewFilter.getStudyIds()));
 
         assertEquals(4, clinicalEventTypeCounts.size());
 
@@ -55,7 +55,7 @@ public void getClinicalEventTypeCounts() {
         dataFilter.setValues(List.of(dataFilterValue));
         studyViewFilter.setClinicalEventFilters(List.of(dataFilter));
 
-        clinicalEventTypeCounts = studyViewMapper.getClinicalEventTypeCounts(StudyViewFilterHelper.build(studyViewFilter, null, null));
+        clinicalEventTypeCounts = studyViewMapper.getClinicalEventTypeCounts(StudyViewFilterHelper.build(studyViewFilter, null, null, studyViewFilter.getStudyIds()));
 
         assertEquals(3, clinicalEventTypeCounts.size());