fix clinical data NA counts for multiple studies

cBioPortal · Sep 5, 2024 · d791618 · d791618
1 parent b36bffb
commit d791618
Show file tree

Hide file tree

Showing 5 changed files with 67 additions and 2 deletions.
diff --git a/src/main/java/org/cbioportal/persistence/StudyViewRepository.java b/src/main/java/org/cbioportal/persistence/StudyViewRepository.java
@@ -46,6 +46,8 @@ public interface StudyViewRepository {
 
     int getFilteredSamplesCount(StudyViewFilterContext studyViewFilterContext);
 
+    int getFilteredPatientCount(StudyViewFilterContext studyViewFilterContext);
+
     Map<String, Set<String>> getMatchingGenePanelIds(StudyViewFilterContext studyViewFilterContext, String alterationType);
 
     int getTotalProfiledCountsByAlterationType(StudyViewFilterContext studyViewFilterContext, String alterationType);

diff --git a/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.java b/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.java
@@ -47,6 +47,8 @@ public interface StudyViewMapper {
 
     int getFilteredSamplesCount(@Param("studyViewFilterHelper") StudyViewFilterHelper studyViewFilterHelper);
 
+    int getFilteredPatientsCount(@Param("studyViewFilterHelper") StudyViewFilterHelper studyViewFilterHelper);
+
     List<GenePanelToGene> getMatchingGenePanelIds(StudyViewFilterHelper studyViewFilterHelper, String alterationType);
 
     int getTotalProfiledCountByAlterationType(StudyViewFilterHelper studyViewFilterHelper, String alterationType);

diff --git a/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMyBatisRepository.java b/src/main/java/org/cbioportal/persistence/mybatisclickhouse/StudyViewMyBatisRepository.java
@@ -156,6 +156,11 @@ public int getFilteredSamplesCount(StudyViewFilterContext studyViewFilterContext
         return mapper.getFilteredSamplesCount(createStudyViewFilterHelper(studyViewFilterContext));
     }
 
+    @Override
+    public int getFilteredPatientCount(StudyViewFilterContext studyViewFilterContext) {
+        return mapper.getFilteredPatientsCount(createStudyViewFilterHelper(studyViewFilterContext));
+    }
+
     @Override
     public Map<String, Set<String>> getMatchingGenePanelIds(StudyViewFilterContext studyViewFilterContext, String alterationType) {
         return mapper.getMatchingGenePanelIds(createStudyViewFilterHelper(studyViewFilterContext), alterationType)

diff --git a/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java b/src/main/java/org/cbioportal/service/impl/StudyViewColumnarServiceImpl.java
@@ -97,14 +97,60 @@ public Map<String, ClinicalDataType> getClinicalAttributeDatatypeMap() {
 
     @Override
     public List<ClinicalDataCountItem> getClinicalDataCounts(StudyViewFilter studyViewFilter, List<String> filteredAttributes) {
-        return studyViewRepository.getClinicalDataCounts(createContext(studyViewFilter), filteredAttributes)
+        Map<String, ClinicalDataType> clinicalAttributeDatatypeMap = this.getClinicalAttributeDatatypeMap();
+        StudyViewFilterContext studyViewFilterContext = createContext(studyViewFilter);
+
+        int filteredSamplesCount = studyViewRepository.getFilteredSamplesCount(studyViewFilterContext);
+        int filteredPatientsCount = studyViewRepository.getFilteredPatientCount(studyViewFilterContext);
+
+        List<ClinicalDataCountItem> clinicalDataCountItems = studyViewRepository
+            .getClinicalDataCounts(createContext(studyViewFilter), filteredAttributes)
             .stream().collect(Collectors.groupingBy(ClinicalDataCount::getAttributeId))
             .entrySet().parallelStream().map(e -> {
                 ClinicalDataCountItem item = new ClinicalDataCountItem();
                 item.setAttributeId(e.getKey());
                 item.setCounts(e.getValue());
                 return item;
-            }).collect(Collectors.toList());
+            }).toList();
+
+        // This is postprocessing to adjust NA counts for multiple studies
+        for (ClinicalDataCountItem clinicalDataCountItem: clinicalDataCountItems) {
+            Integer totalClinicalDataCount = clinicalDataCountItem
+                .getCounts()
+                .stream()
+                .map(ClinicalDataCount::getCount)
+                .reduce(0, Integer::sum);
+            // depending on clinical data type we either use filtered sample count or filtered patient count
+            int filteredCount = clinicalAttributeDatatypeMap.get(clinicalDataCountItem.getAttributeId()) == ClinicalDataType.SAMPLE ?
+                filteredSamplesCount: filteredPatientsCount;
+            int casesWithoutClinicalData = filteredCount - totalClinicalDataCount;
+
+            if (casesWithoutClinicalData > 0) {
+                // find "NA"
+                ClinicalDataCount naClinicalDataCount = clinicalDataCountItem
+                    .getCounts()
+                    .stream()
+                    .filter(c -> c.getValue().equals("NA"))
+                    .findFirst()
+                    .orElse(null);
+
+                if (naClinicalDataCount != null) {
+                    // if there is already "NA" count add cases without clinical data to the existing count
+                    naClinicalDataCount.setCount(naClinicalDataCount.getCount() + casesWithoutClinicalData);
+                }
+                else {
+                    // if no NA count, create a new one
+                    // this should only happen when there are multiple studies
+                    naClinicalDataCount = new ClinicalDataCount();
+                    naClinicalDataCount.setAttributeId(clinicalDataCountItem.getAttributeId());
+                    naClinicalDataCount.setValue("NA");
+                    naClinicalDataCount.setCount(casesWithoutClinicalData);
+                    clinicalDataCountItem.getCounts().add(naClinicalDataCount);
+                }
+            }
+        }
+
+        return clinicalDataCountItems;
     }
 
     @Override

diff --git a/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml b/src/main/resources/org/cbioportal/persistence/mybatisclickhouse/StudyViewMapper.xml
@@ -227,6 +227,12 @@
         WHERE sample_unique_id IN (<include refid="sampleUniqueIdsFromStudyViewFilter"/>)        
     </sql>
 
+    <sql id="getTotalPatientCount">
+        SELECT count(distinct patient_unique_id) as count
+        FROM sample_derived sd
+        WHERE sample_unique_id IN (<include refid="sampleUniqueIdsFromStudyViewFilter"/>)
+    </sql>
+
     <!-- for /genomic-data-counts/fetch - (returns GenomicDataCountItem objects) -->
     <select id="getCNACounts" resultMap="GenomicDataCountItemResultMap">
         <bind name="profileType" value="genomicDataFilters[0].profileType" />
@@ -397,6 +403,10 @@
         <include refid="getTotalSampleCount"/>
     </select>
 
+    <select id="getFilteredPatientsCount" resultType="int">
+        <include refid="getTotalPatientCount"/>
+    </select>
+
     <!-- Get Matching Gene Panel Ids for all Samples after study view filter -->
     <select id="getMatchingGenePanelIds" resultType="org.cbioportal.model.GenePanelToGene">
         SELECT